Respiratory pressure measurement method and system

Abstract

The present application provides a respiratory pressure measurement method and system, belonging to the field of human physiological signal monitoring, processing the respiratory pressure waveform to obtain a candidate respiratory segment conforming to the respiratory rhythm; calculating the amplitude of each candidate respiratory segment, and establishing an amplitude sliding history queue based on the historical amplitude, calculating the dynamic average amplitude of the amplitude sliding history queue, judging the amplitude validity of the current candidate respiratory segment based on the dynamic average amplitude to obtain an amplitude qualified candidate segment; calculating the normalized cross-correlation coefficient of the rising edge waveform sequence after amplitude normalization and the standard half-sine wave template, and determining the amplitude qualified candidate segment with the normalized cross-correlation coefficient greater than or equal to the morphological matching threshold as an effective respiratory segment; taking the median in the baseline sequence as a dynamic reference zero point to obtain a respiratory pressure display value; and outputting a respiratory abnormality alarm signal through the calculation of the instantaneous average respiratory rate combined with the alarm triggering condition; improving the accuracy and anti-interference ability of respiratory detection.

Description

Technical Field

[0001] This invention relates to the field of human physiological signal monitoring technology, and in particular to a method and system for measuring respiratory pressure. Background Technology

[0002] Respiratory rate, one of the four vital signs of the human body, has significant assessment value in clinical monitoring, sleep monitoring, rehabilitation training, and daily health management. Traditional respiratory monitoring methods mainly include impedance methods, thermal sensor methods, carbon dioxide concentration detection methods, and chest and abdominal belt respiratory sensors. However, these methods generally suffer from problems such as sensitivity to electromagnetic interference, poor wearing comfort, and insufficient long-term monitoring stability. In recent years, fiber optic sensing technology has been gradually introduced into the field of human respiratory signal acquisition due to its advantages such as resistance to electromagnetic interference, small size, light weight, high sensitivity, and intrinsic safety. Fiber optic sensors can convert the chest and abdominal deformation, pressure changes, or vibration signals caused by respiration into changes in light intensity, wavelength, or phase, thereby achieving continuous and real-time respiratory monitoring.

[0003] In existing fiber optic sensor-based respiratory monitoring technologies, a common approach is to indirectly reflect respiratory activity through changes in light intensity caused by fiber deformation. For example, Chinese invention patent application CN112190255A, entitled "A Wearable Respiratory Monitoring Device Based on D-Type Plastic Fiber Optic," includes: a light source, plastic fiber optic cable, a fixing strap, a fixing block, a fixing plate, a photodetector module, a signal acquisition module, and a terminal. The light source is connected to the input end of the photodetector module via the plastic fiber optic cable; the output end of the photodetector module is connected to the input end of the signal acquisition module; and the output end of the signal acquisition module is connected to the terminal. The plastic fiber optic cable includes a sensing area, which is the area where the plastic fiber undergoes a side-throwing process with a throw length of 1cm-6cm and a throw depth of 0.2m. The sensor area is 0.8mm in diameter and fixed to a transparent plastic plate. Both the fiber optic inlet and outlet of the plastic plate are secured with fixing blocks made of solidified AB glue. These fixing blocks are sewn onto a stretchable elastic band, which is secured to the user's abdomen via Velcro at both ends, positioning the fixing blocks in the center of the abdomen. During operation, the sensor area bends and deforms according to the abdominal deformation caused by the user's breathing, increasing light emission and thus changing the voltage output of the photodetector module. The signal acquisition module transmits the voltage change to the terminal (computer). The terminal uses wavelet decomposition to eliminate baseline drift and interference from the received raw signal and performs Fourier transform on the wavelet-decomposed signal to obtain a spectrum. Finally, the user's respiratory rate is monitored based on the voltage change. This respiratory monitoring device, through its hardware-level fiber optic sensing structure design combined with wavelet decomposition and Fourier transform signal processing, achieves the basic functions of fiber optic respiratory monitoring, representing a typical implementation path in this field.

[0004] Although the existing technical solutions, represented by the Chinese invention patent application with publication number CN112190255A, have achieved the basic function of respiratory monitoring based on fiber optic sensors, there are still several significant defects in practical applications: (1) The terminal signal processing of this solution only uses wavelet decomposition to eliminate baseline drift and interference, and Fourier transform to obtain the spectrum. This method is essentially a global frequency domain analysis method, which is difficult to identify the start and end times and peak positions of each respiratory waveform in real time and accurately. Especially when the respiratory waveform is affected by non-respiratory sources such as coughing, body position changes, clothing friction or heartbeat vibration, wavelet decomposition and Fourier transform cannot effectively distinguish between the real respiratory waveform and the interference waveform, and it is easy to misjudge the interference as breathing or cause the real breathing to be missed. (2) This solution lacks an adaptive amplitude verification mechanism. Since the respiratory depth of different individuals varies significantly, and the respiratory amplitude of the same individual will also change greatly in different sleep stages or exercise states, fixed thresholds or single wavelet analysis methods cannot dynamically adapt to such amplitude fluctuations, resulting in decreased detection sensitivity or increased false detection rate.

[0005] In summary, existing technologies urgently need a respiratory detection algorithm that can integrate adaptive amplitude verification, waveform morphology quantization matching, and multi-timescale respiratory anomaly judgment to overcome the above-mentioned shortcomings. Summary of the Invention

[0006] The technical problem to be solved by this invention is: how to improve the accuracy and anti-interference ability of respiratory detection.

[0007] This invention solves the above-mentioned technical problems through the following technical solution: a respiratory pressure measurement method, comprising:

[0008] The respiratory pressure waveform is processed to obtain candidate respiratory segments that conform to the respiratory rhythm; The amplitude of each candidate respiratory segment is calculated, and a sliding historical amplitude queue is established based on the historical amplitude. The dynamic average amplitude of the sliding historical amplitude queue is calculated, and the validity of the current candidate respiratory segment is judged based on the dynamic average amplitude to obtain the candidate segment with qualified amplitude. The waveform of the actual respiratory rising segment in the amplitude qualified candidate segment is normalized. The normalized cross-correlation coefficient between the normalized rising edge waveform sequence and the standard half-sine wave template is calculated. The amplitude qualified candidate segment with the normalized cross-correlation coefficient greater than or equal to the morphological matching threshold is determined as a valid respiratory segment. The trough pressure values ​​of multiple effective respiratory segments are continuously recorded and a baseline sequence is formed. The median of the baseline sequence is used as the dynamic reference zero point. The respiratory pressure display value is obtained by subtracting the dynamic reference zero point from the collected real-time pressure data. The instantaneous average respiratory rate of the effective respiratory segments within the short-term and long-term respiratory rate sliding windows is calculated, and an abnormal respiratory alarm signal is output in combination with the alarm triggering conditions.

[0009] Beneficial effects: This invention achieves high-precision, high-robustness, and low-false-alarm-rate respiratory waveform identification and respiratory event judgment by integrating adaptive amplitude history queue verification, half-sine waveform morphology matching based on normalized cross-correlation coefficients, dynamic baseline calibration, and dual-window respiratory rate anomaly monitoring. Specifically, this invention constructs a two-level verification architecture of initial amplitude screening and final morphology judgment. The dynamic sliding queue established based on historical effective respiratory amplitudes calculates the dynamic average amplitude and judges the validity of the current candidate respiratory segment based on the dynamic average amplitude. This can effectively eliminate amplitude abnormal segments such as instantaneous pressure spikes and extremely weak respiratory fluctuations, while adapting to the differences in respiratory depth among different individuals. For candidate segments that meet the amplitude criteria, this invention normalizes the amplitude of the actual respiratory rise waveform (from trough to peak) and calculates a normalized cross-correlation coefficient with a standard half-sine wave template. Candidate segments with a normalized cross-correlation coefficient greater than or equal to the morphological matching threshold are considered valid respiratory segments. The morphological design of the half-sine template conforms to the physiological characteristics of the natural human respiratory pressure waveform: "slow rise, smooth curve, and symmetrical shape." The normalized cross-correlation calculation eliminates interference from amplitude scale and mean offset, retaining only the waveform contour morphological information for similarity measurement. A baseline sequence is constructed using the trough pressure values ​​of multiple consecutive valid respiratory segments, and the median is calculated as a dynamic reference zero point, effectively avoiding interference from single waveform anomalies on baseline calibration.

[0010] Preferably, the process of processing the respiratory pressure waveform to obtain candidate respiratory segments that conform to the respiratory rhythm includes: According to the normal human respiratory rate range, the potential respiratory waveform components are separated from the respiratory pressure waveform, and the first-order difference of the potential respiratory waveform components is calculated to obtain the first-order derivative sequence. Based on the derivative zero-crossing detection rule, the rising and falling inflection points of the respiratory pressure waveform are initially located. Based on the physiological constraints of a single human respiratory cycle, the time interval between adjacent rising inflection points is screened, and time segments within the physiological constraints of a single human respiratory cycle are marked as candidate respiratory segments that conform to the respiratory rhythm.

[0011] Preferably, the normal human respiratory rate range is 0.1Hz to 0.8Hz; the physiological constraint of a single human respiratory cycle is 1.2 seconds to 5 seconds; time segments with a time interval of less than 1.2 seconds between adjacent rising edge inflection points are judged as high-frequency interference and discarded; time segments between 1.2 seconds and 5 seconds are marked as candidate respiratory segments that conform to the respiratory rhythm; and time segments with a time interval of more than 5 seconds between adjacent rising edge inflection points are temporarily stored in the observation buffer and a no-breathing timer is started.

[0012] Beneficial effects: This invention combines the normal human respiratory rate range (0.1Hz-0.8Hz) with the detection of zero crossing of the first-order differential derivative to initially locate waveform inflection points. Based on the physiological constraints of a single respiratory cycle (1.2 seconds to 5 seconds), candidate segments are screened, discarding high-frequency interference that is too short (<1.2 seconds) and invalid segments that are too long (>5 seconds). Simultaneously, an observation buffer and a non-respiratory timer are introduced. By effectively filtering out non-respiratory interference such as heartbeat vibrations and body tremors from a physiological perspective, high-quality candidate waveform data is provided for subsequent high-precision identification.

[0013] Preferably, the amplitude of each candidate respiratory segment is calculated as follows: calculate the global maximum value and global minimum value of the respiratory pressure waveform within each candidate respiratory segment, and subtract the global minimum value from the global maximum value to obtain the amplitude of each candidate respiratory segment.

[0014] Preferably, the method for determining the amplitude validity of the current candidate respiratory segment based on the dynamic average amplitude is as follows: candidate respiratory segments whose amplitudes meet the validity judgment criteria are determined as qualified amplitude candidate segments. The validity judgment criteria are: , For amplitude, This is the dynamic average amplitude.

[0015] Beneficial effects: By establishing an amplitude sliding history queue to store the amplitude data of multiple recent effective respiratory segments and dynamically calculating their average value, adaptive judgment of the amplitude of the current candidate respiratory segment is achieved. The validity judgment condition is as follows: By implementing the above settings, spurious segments such as instantaneous pressure spikes and weak respiratory fluctuations are effectively eliminated, and the problem of large amplitude fluctuations caused by changes in breathing depth and mask loosening is solved, significantly improving the robustness and adaptability of respiratory detection.

[0016] Preferably, the method for normalizing the amplitude of the actual respiratory rise waveform in the candidate segments with acceptable amplitude is as follows:

[0017] in, This is the rising edge waveform sequence after amplitude normalization. The rising edge of the candidate segment with acceptable amplitude is the original pressure sampling sequence. , This represents the total number of sampling points during the ascending respiratory rate. , These are the minimum and maximum pressure values ​​within the actual respiratory rise waveform of the candidate segments with acceptable amplitude, respectively.

[0018] Beneficial effects: By using the maximum and minimum values ​​of the rising edge of a single breath as the scaling reference, the original pressure sampling points are linearly mapped to eliminate the amplitude scale differences caused by different breathing depths and wearing conditions, and only the waveform contour morphology features are retained. This ensures that the actual collected waveform and the standard half-sine template have the same data range, providing unified calculation conditions for subsequent waveform similarity measurement and matching.

[0019] Preferably, the method for calculating the normalized cross-correlation coefficient between the amplitude-normalized rising edge waveform sequence and the standard half-sine wave template is as follows:

[0020] in, To normalize the cross-correlation coefficient, This is the rising edge waveform sequence after amplitude normalization. The rising edge waveform sequence after amplitude normalization The overall average, These are sampled values ​​of a standard half-sine wave template waveform. Sampled values ​​of standard half-sine wave template waveform The overall average, , This represents the total number of sampling points during the ascending respiratory rate. The standard half-sine wave template is:

[0021] in, This is a standard half-sine wave template waveform. For the real-time sampling time variable during the fragment's ascent process, , This represents the total number of sampling points during the ascending respiratory rate. It is the circumference constant. .

[0022] Beneficial effects: By performing mean removal, point-by-point correlation calculations, and standardization on the normalized actual rising waveform and the standard half-sine template waveform, the contour similarity of the two waveforms is quantified; computational interference caused by mean offset and amplitude scale is eliminated, retaining only the differences in curve morphology features, and completing the respiratory waveform morphology verification in the form of quantized coefficients, achieving accurate filtering of irregular interference waveforms. The construction of the standard half-sine template, constrained by the sampling length of the respiratory rising segment, generates a smooth and continuous half-sine standard waveform, which serves as a reference template for respiratory morphology, used to quantify and compare the waveform features of actual candidate respiratory segments, and distinguish between normal breathing and non-respiratory interference waveforms.

[0023] Preferably, the morphological matching threshold is 0.85. After determining the qualified candidate segments with normalized cross-correlation coefficients greater than or equal to the morphological matching threshold as valid respiratory segments, the amplitude of the valid respiratory segments is confirmed as valid amplitude, and the valid amplitude is pushed into the amplitude sliding history queue. The amplitude sliding history queue automatically pops the earliest stored data to maintain the constant length of the amplitude sliding history queue, and at the same time outputs the valid peak time and peak pressure value of the valid respiratory segment.

[0024] Beneficial Effects: This invention collects real-world data from multiple scenarios, including various raw pressure waveform samples such as normal steady breathing, shallow breathing, deep breathing, body swaying, body shaking, short-term coughing, clothing friction, and low-frequency equipment vibration. Normalized cross-correlation coefficients (RCCs) between each waveform and a standard half-sine breathing template are calculated. The RCCs for the rising phase of real breathing range from 0.85 to 0.97, those for slight body movement disturbances range from 0.70 to 0.84, while those for heartbeat vibrations, coughing, and shaking are below 0.70. By setting a threshold of 0.85, it is possible to accurately distinguish real breathing from various non-respiratory disturbances at the morphological level, overcoming the shortcomings of traditional methods that rely solely on amplitude or frequency thresholds and easily misjudge coughing, body position changes, etc., as breathing.

[0025] Preferably, the median in the baseline sequence is calculated as follows: Sort the trough pressure values ​​of an odd number of effective breathing segments in ascending order:

[0026] , ... ... This represents the trough pressure values ​​of the odd-numbered effective respiratory segments in the baseline sequence. It is a positive integer greater than or equal to 1; the trough pressure value As the median in the baseline sequence.

[0027] Preferably, during the continuous recording of trough pressure values ​​of multiple effective respiratory segments, when the time without effective respiratory peaks exceeds the no-breathing alarm threshold, the baseline sequence update is stopped and the dynamic reference zero point is frozen, keeping it as the dynamic reference zero point calculated from the most recent effective respiratory cycle; when two consecutive effective respiratory peaks are detected, the first effective respiratory peak is used to restart the baseline recording mechanism but not immediately update the dynamic reference zero point, and the second effective respiratory peak is used to confirm that the respiratory rhythm is stable and then automatically restore the baseline sequence rolling update mechanism, re-acquire the most recent inspiratory start baseline, and calculate the median.

[0028] Beneficial effects: This invention supports real-time offset zeroing display and one-click zeroing command from the user. More importantly, when the system detects that consecutive invalid respiratory peaks exceed a preset threshold, it automatically freezes the dynamic baseline update to prevent zero-point drift caused by sensor noise accumulation during apnea. After breathing resumes, a gradual strategy of "restarting recording with the first recovery peak and resuming rolling updates with the second recovery peak" is used to correct the drift error accumulated during the freeze period in one go. This mechanism ensures the accuracy and stability of pressure measurement during long-term continuous monitoring and solves the pain point of detection failure caused by zero-point shift after long-term operation of traditional methods.

[0029] Preferably, the alarm triggering conditions include bradyventricular breathing alarm triggering conditions and apnea alarm triggering conditions. The instantaneous average respiratory rate of the effective respiratory segment within the short-term respiratory rate sliding window and the long-term respiratory rate sliding window are calculated respectively to obtain the short-term average respiratory rate and the long-term average respiratory rate. When both the short-term average respiratory rate and the long-term average respiratory rate are lower than [a certain value] per minute... Once, and lasting longer than A short window period is required to meet the conditions for triggering the bradybreathing alarm. Indicates the number of breaths. This indicates the number of short-term window periods. When the time without a valid peak marker exceeds the no-breathing alarm threshold, and there are no candidate breathing segments that meet the amplitude validity judgment conditions during the no-breathing alarm threshold period, the apnea alarm trigger condition is met.

[0030] The present invention also provides a respiratory pressure measurement system, comprising: The waveform preprocessing module is used to process the respiratory pressure waveform to obtain candidate respiratory segments that conform to the respiratory rhythm; The adaptive amplitude verification module is used to calculate the amplitude of each candidate respiratory segment, establish an amplitude sliding history queue based on historical amplitudes, calculate the dynamic average amplitude of the amplitude sliding history queue, and judge the amplitude validity of the current candidate respiratory segment based on the dynamic average amplitude to obtain the candidate segment with qualified amplitude. The waveform morphology matching verification module is used to normalize the amplitude of the actual respiratory rising segment waveform in the amplitude qualified candidate segment, calculate the normalized cross-correlation coefficient between the normalized rising edge waveform sequence and the standard half-sine wave template, and determine the amplitude qualified candidate segment with the normalized cross-correlation coefficient greater than or equal to the morphology matching threshold as a valid respiratory segment. The dynamic baseline calibration module is used to continuously record the trough pressure values ​​of multiple effective respiratory segments and form a baseline sequence. The median in the baseline sequence is used as the dynamic reference zero point. The respiratory pressure display value is obtained by subtracting the dynamic reference zero point from the collected real-time pressure data. The dual-window respiratory rate abnormality monitoring module is used to calculate the instantaneous average respiratory rate of the effective respiratory segment within the short-term respiratory rate sliding window and the long-term respiratory rate sliding window, and output a respiratory abnormality alarm signal in combination with the alarm triggering conditions. Attached Figure Description

[0031] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0032] Figure 1 This is a flowchart of the respiratory pressure measurement method provided in Embodiment 1 of the present invention; Figure 2 This is a flowchart of obtaining a qualified candidate segment of amplitude in the respiratory pressure measurement method provided in Embodiment 1 of the present invention; Figure 3 This is a flowchart of the effective respiratory segment determination in the respiratory pressure measurement method provided in Embodiment 1 of the present invention; Figure 4 This is a sequence diagram of the baseline sequence formed by continuously recording the trough pressure values ​​of five effective respiratory segments in the respiratory pressure measurement method provided in Embodiment 1 of the present invention; Figure 5 This is a flowchart of the respiratory rate monitoring and abnormal alarm determination in the respiratory pressure measurement method provided in Embodiment 1 of the present invention; Figure 6 This is a schematic diagram of the respiratory pressure measurement system provided in Embodiment 2 of the present invention.

[0033] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0035] Example 1 like Figure 1 As shown, this embodiment provides a method for measuring respiratory pressure, including the following steps: Step 1: Process the respiratory pressure waveform to obtain candidate respiratory segments that conform to the respiratory rhythm.

[0036] Before processing the respiratory pressure waveform, the raw pressure data output by the fiber optic demodulator is first acquired in real time via User Datagram Protocol (UDP) communication and formed into a continuous waveform sequence. This raw pressure data is measured by a fiber optic sensor. A sliding time window is used to analyze the local characteristics of the signal. Noise filtering and baseline correction are completed by combining low-pass filtering with dynamically adjusted cutoff frequency and morphological opening operations. Simultaneously, an adaptive threshold mechanism is introduced to handle slow drift caused by loose breathing equipment or changes in body position, outputting a stable pressure waveform without high-frequency noise or baseline shift. The obtained stable pressure waveform is the respiratory pressure waveform.

[0037] Step 1 involves processing the respiratory pressure waveform to obtain candidate respiratory segments that match the respiratory rhythm. Step 1.1: Separate the potential respiratory waveform components from the respiratory pressure waveform according to the normal human respiratory frequency range, calculate the first-order difference of the potential respiratory waveform components, and obtain the first-order derivative sequence.

[0038] Step 1.2: Based on the derivative zero-crossing detection rule, preliminarily locate the rising edge inflection point and the falling edge inflection point of the respiratory pressure waveform.

[0039] Step 1.3: Based on the physiological constraints of a single human respiratory cycle, filter the time interval between adjacent rising edge inflection points, and mark the time segments within the physiological constraints of a single human respiratory cycle as candidate respiratory segments that conform to the respiratory rhythm.

[0040] In this invention, the normal human respiratory rate ranges from 0.1Hz to 0.8Hz. A 0.1Hz to 0.8Hz bandpass filter is used to perform secondary filtering on the respiratory pressure waveform to separate potential respiratory waveform components. The physiological constraint for a single human respiratory cycle is 1.2 seconds to 5 seconds. Time segments with a time interval of less than 1.2 seconds between adjacent rising edge inflection points are identified as high-frequency interference and discarded. Time segments between 1.2 seconds and 5 seconds are marked as candidate respiratory segments conforming to the respiratory rhythm. Time segments with a time interval of more than 5 seconds between adjacent rising edge inflection points are temporarily stored in an observation buffer, and a no-respiration timer is started.

[0041] Step 1 of this invention combines the normal human respiratory frequency range (0.1Hz-0.8Hz) and the zero-crossing detection of the first-order differential derivative to initially locate the waveform inflection point. Based on the physiological constraints of a single respiratory cycle (1.2 seconds to 5 seconds), candidate respiratory segments are screened, discarding high-frequency interference that is too short (<1.2 seconds) and invalid segments that are too long (>5 seconds). Simultaneously, an observation buffer and a non-respiratory timer are introduced. By effectively filtering out non-respiratory interference waveforms such as heartbeat vibrations and body tremors from a physiological perspective, only candidate respiratory segments that conform to the respiratory rhythm are retained for subsequent identification stages, providing high-quality candidate waveform data for subsequent high-precision identification. Candidate respiratory segments that conform to the respiratory rhythm are retained and sent to the subsequent identification stage. Specifically, they are temporarily stored in a structured form in the system's memory temporary buffer as a sequence of waveform sampling points, start and end timestamps, and segment duration. This serves as the direct input for amplitude verification and waveform morphology matching in Step 2. Interference segments that do not conform to physiological constraints are directly discarded and not stored.

[0042] Here's an example: Suppose that after processing in step 1.2, the timestamps of two adjacent rising edge inflection points are located at 3.10 seconds and 5.50 seconds respectively. The time interval between them is 2.10 seconds, which falls within the physiological constraint range of a single human respiratory cycle (1.2 seconds to 5 seconds). The corresponding waveform segment (time segment) is a candidate respiratory segment that conforms to the respiratory rhythm. A candidate respiratory segment that conforms to the respiratory rhythm contains a continuous sequence of pressure sampling values, such as 126, 142, 168, 185, 171, 153, and 130. The length of the time segment is 2.40 seconds, and the number of sampling points corresponds to the system sampling rate. This segment will be directly sent to step 2 for amplitude calculation and waveform morphology matching.

[0043] Step 2: Calculate the amplitude of each candidate respiratory segment, establish an amplitude sliding historical queue based on historical amplitudes, calculate the dynamic average amplitude of the amplitude sliding historical queue, and judge the amplitude validity of the current candidate respiratory segment based on the dynamic average amplitude to obtain qualified amplitude candidate segments.

[0044] See Figure 2 In step 2, the amplitude of each candidate respiratory segment is calculated as follows: The global maximum and global minimum values ​​of the respiratory pressure waveform within each candidate respiratory segment are calculated. The global minimum value is then subtracted from the global maximum value to obtain the amplitude of each candidate respiratory segment. The calculation formula is expressed as follows:

[0045] in, The amplitude of the candidate respiratory segment. This represents the global maximum value of the pressure sampled values ​​within a candidate respiratory segment. This represents the global minimum value of the pressure sampling value within a candidate respiratory segment. This represents the pressure sampling value within the candidate respiratory segment.

[0046] This invention establishes an amplitude sliding history queue on a temporary buffer within the system. Store the most recently completed Amplitude data of each effective respiratory segment , , Indicates the amplitude sliding history queue The total number of valid respiratory segments stored in the system, in this embodiment The possible values ​​are 10, therefore, .

[0047] Dynamic average amplitude of the amplitude sliding history queue The calculation method is as follows:

[0048] Simultaneously, the standard deviation of the amplitude can be calculated. :

[0049] when When the value is 10, the dynamic average amplitude of the amplitude sliding historical queue The calculation method is as follows:

[0050] Simultaneously, the standard deviation of the amplitude can be calculated. :

[0051] Among them, the dynamic average amplitude of the amplitude sliding historical queue This will be used to provide an adaptive amplitude interpretation benchmark for subsequent new segments.

[0052] The method for determining the validity of current candidate respiratory segments based on dynamic average amplitude is as follows: Candidate respiratory segments whose amplitudes meet the validity judgment criteria are identified as qualified candidate segments. The validity judgment criteria are: , For amplitude, This is the dynamic average amplitude.

[0053] The above validity criteria ensure that only respiratory segments with amplitude values ​​no less than 0.4 times and no more than 3.0 times the historical average amplitude can proceed to the subsequent waveform morphology matching stage. This eliminates false segments such as instantaneous pressure spikes and weak respiratory fluctuations, resolving the issue of large amplitude fluctuations caused by changes in breathing depth or mask loosening, and significantly improving the robustness and adaptability of respiratory detection. For candidate respiratory segments that pass the amplitude criteria, step 3, waveform morphology matching verification, is then performed.

[0054] Step 3, see Figure 3 The amplitude of the actual respiratory rise segment waveform in the amplitude qualified candidate segment is normalized. The normalized cross-correlation coefficient between the normalized rise edge waveform sequence and the standard half-sine wave template is calculated. The amplitude qualified candidate segment with the normalized cross-correlation coefficient greater than or equal to the morphological matching threshold is determined as a valid respiratory segment.

[0055] The method for normalizing the amplitude of the actual respiratory rise segment waveform in the candidate segments with qualified amplitude in step 3 is as follows:

[0056] in, This is the rising edge waveform sequence after amplitude normalization. The rising edge of the candidate segment with acceptable amplitude is the original pressure sampling sequence. , This represents the total number of sampling points during the ascending respiratory rate. , These represent the minimum and maximum pressure values ​​within the actual respiratory rise waveform of the amplitude-qualified candidate segment, respectively. For example, suppose the minimum pressure value within the actual respiratory rise waveform of the amplitude-qualified candidate segment is 20, the maximum pressure value is 80, and the original pressure sampling sequence of the rise edge of the amplitude-qualified candidate segment has a value of 50 at a certain point. Then, the pressure value after amplitude normalization at that point is... It is 0.5, and the calculation method is as follows:

[0057] By performing amplitude normalization processing on the actual respiratory rise segment waveform in the candidate segments with qualified amplitude, the amplitude range of the actual respiratory rise segment waveform is uniformly mapped to an interval. .

[0058] Before calculating the normalized cross-correlation coefficient between the amplitude-normalized rising edge waveform sequence and the standard half-sine wave template in step 3, the standard half-sine wave template is first constructed. The standard half-sine wave template is as follows:

[0059] in, This is a standard half-sine wave template waveform. For the real-time sampling time variable during the fragment's ascent process, , Starting from 0 and incrementing, This represents the total number of sampling points during the ascending respiratory rate. It is the circumference constant. , Used to constrain the interval of a sine curve. , It generates a single-segment rising half-sine curve that matches the natural breathing pressure waveform of the human body: a slow rise, a smooth curve, and a symmetrical shape, which is the standard ideal shape of the rising edge of breathing.

[0060] The standard half-sine wave template is constructed by using the sampling length of the rising respiratory segment as a constraint to generate a smooth and continuous standard half-sine waveform, which serves as a reference template for respiratory morphology. This template is used to quantify and compare the waveform characteristics of actual candidate respiratory segments and distinguish between normal breathing and non-breathing interference waveforms.

[0061] By using the maximum and minimum values ​​of the rising edge of a single breath segment as the scaling reference, the original pressure sampling points are linearly mapped to eliminate the amplitude scale differences caused by different breathing depths and wearing conditions. Only the waveform contour morphology features are retained, ensuring that the actual acquired waveform and the standard half-sine template have the same data range, providing unified calculation conditions for subsequent waveform similarity measurement and matching.

[0062] The method for calculating the normalized cross-correlation coefficient between the amplitude-normalized rising edge waveform sequence and the standard half-sine wave template in step 3 is as follows:

[0063] in, The normalized cross-correlation coefficient is used to characterize the similarity between the actual normalized rising waveform and the standard half-sine template waveform. The range of values ​​for is: , The closer the value is to 1, the more consistent the waveform shape. This is the rising edge waveform sequence after amplitude normalization. The rising edge waveform sequence after amplitude normalization The overall average, These are sampled values ​​of a standard half-sine wave template waveform. Sampled values ​​of standard half-sine wave template waveform The overall average, , This represents the total number of sampling points during the ascending respiratory tract.

[0064] Among them, the rising edge waveform sequence after amplitude normalization Overall average The calculation method is as follows:

[0065] Standard half-sine wave template waveform sampling value Overall average The calculation method is as follows:

[0066] Step 3 involves performing mean removal, point-by-point correlation calculations, and standardization on the normalized actual rising waveform and the standard sine template waveform to quantify the contour similarity of the two waveforms, eliminate calculation interference caused by mean offset and amplitude scale, retain only the differences in curve shape characteristics, and complete the breathing waveform shape verification in the form of quantization coefficients to achieve accurate filtering of irregular interference waveforms.

[0067] In this embodiment, the morphological matching threshold is 0.85. For amplitude-qualified candidate segments that meet the amplitude validity judgment conditions but whose normalized cross-correlation coefficient is less than or equal to the morphological matching threshold, this invention determines them as interference signals and removes them. After completing the effective respiratory peak marking, data update is performed: after determining the amplitude-qualified candidate segments with normalized cross-correlation coefficients greater than or equal to the morphological matching threshold as effective respiratory segments, the amplitude of the effective respiratory segments is confirmed as the effective amplitude. and the effective amplitude Push into amplitude sliding history queue The amplitude sliding history queue automatically pops the earliest stored data to maintain a constant length. It also outputs the effective peak time and peak pressure value of the effective respiratory segment.

[0068] This invention sets the morphological matching threshold at 0.85. Specifically, it collects real data from multiple scenarios, including various raw pressure waveform samples such as normal steady breathing, shallow breathing, deep breathing, body swaying, body shaking, short-term coughing, clothing friction, and low-frequency equipment vibration. The normalized cross-correlation coefficients between each waveform and the standard half-sine breathing template are calculated, where: The rising phase of effective real human respiration: normalized cross-correlation coefficient between waveform and standard half-sine respiration template. Slight body movement, minor disturbance: Heartbeat vibration, trembling, coughing, squeezing interference: By using the sample distribution boundary, the critical dividing value between effective breathing and interference signals was selected, and 0.85 was determined as the optimal dividing point.

[0069] Current effective amplitude Push into amplitude sliding history queue The earliest data is popped up. In this embodiment, the queue length is kept at 10. At the same time, the effective peak time and peak pressure value are output for dynamic baseline calibration and bias zeroing in step 4.

[0070] Step 4: Continuously record the trough pressure values ​​of multiple effective respiratory segments and form a baseline sequence. Use the median in the baseline sequence as the dynamic reference zero point. Subtract the dynamic reference zero point from the collected real-time pressure data to obtain the respiratory pressure display value. Calculate the instantaneous average respiratory rate of the effective respiratory segments within the short-term and long-term respiratory rate sliding windows, and output a respiratory abnormality alarm signal based on the alarm triggering conditions.

[0071] After each valid respiratory segment is determined in step 3, the effective peak time and peak pressure value corresponding to the valid respiratory segment are output simultaneously, and the pressure value corresponding to the trough in the valid respiratory segment is extracted. As the first The baseline for the onset of inspiration in each respiratory cycle. This indicates the sequence number of the currently completed effective respiratory cycle. The most recent cycle is recorded continuously. M The inspiratory baseline values ​​of each effective respiratory cycle constitute a baseline sequence, see [link to baseline sequence]. Figure 4 In this embodiment M With a value of 5, the baseline sequence can be represented as follows:

[0072] in, This represents the set of baseline troughs for the most recent five consecutive effective respiratory cycles, summarizing the current trough. The pressure values ​​of each effective respiratory cycle and the trough of the previous four consecutive historical cycles are stored in a fixed window to provide continuous historical samples for subsequent baseline mean calculation and drift correction. This reduces the impact of single waveform anomalies on baseline calibration and ensures the long-term stability of the pressure detection baseline.

[0073] Using the median of the baseline sequence as the dynamic reference zero, the median of the baseline sequence is calculated as follows: Sort the trough pressure values ​​of an odd number of effective breathing segments in ascending order:

[0074] , ... ... This represents the trough pressure values ​​of the odd-numbered effective respiratory segments in the baseline sequence. It is a positive integer greater than or equal to 1; the trough pressure value As the median in the baseline sequence.

[0075] In this embodiment, the five elements in the baseline sequence are sorted in ascending order of value to obtain an ordered sequence:

[0076] Select the third element As median Using the median instead of the mean can effectively avoid excessive deviation of the reference zero point caused by a brief jump in the baseline value during a respiratory cycle.

[0077] Subtracting this dynamic reference zero point from the subsequently acquired real-time pressure data yields the displayed pressure value after offset zeroing, thus achieving real-time offset zeroing display of the pressure measurement value. The calculation method is as follows:

[0078] in, This indicates the respiratory pressure display value. This indicates the collected real-time pressure data.

[0079] When a user triggers a one-click zeroing command through the software interface, the system immediately clears the current baseline sequence cache and uses the instantaneous pressure value detected at the time of triggering. As the new baseline zero point, the dynamic reference zero point is forcibly reset, triggering the instantaneous pressure value detected at the moment of triggering. Assigned to dynamic reference zero ,Right now At the same time, the instantaneous pressure value Copy and fill to the baseline sequence M Positions to maintain sequence length M .

[0080] This invention continuously records the inspiratory initiation baseline (trough pressure value) of the most recent 5 effective respiratory cycles and uses the median instead of the mean as the dynamic reference zero point, effectively avoiding the impact of single waveform anomalies on baseline calibration. It also supports real-time offset zeroing display and one-click zeroing command, solving the measurement error problem caused by slow baseline drift during long-term acquisition by fiber optic sensors and ensuring the long-term stability of the pressure detection baseline.

[0081] In step 4, during the continuous recording of trough pressure values ​​for multiple valid respiratory segments, invalid respiratory peaks (interference waveforms) are directly discarded and not counted as valid peaks. The continuous occurrence of invalid respiratory peaks alone does not trigger a condition. Baseline sequence updates are stopped only when no valid peak passing amplitude and morphology verification is found throughout the entire process and the time gap exceeds the no-respiratory alarm threshold. That is, baseline sequence updates stop when the duration of consecutive invalid respiratory peaks exceeds the no-respiratory alarm threshold. In this embodiment, the no-respiratory alarm threshold is 20 seconds. Baseline sequence updates are then stopped, and the dynamic reference zero point is frozen. This keeps it at the dynamic reference zero point calculated from the most recent effective respiratory cycle. This is to prevent zero-point drift caused by the accumulation of sensor noise during apnea.

[0082] After breathing resumes, the system re-detects two consecutive valid respiratory peaks. The first recovered peak is used to restart the baseline recording mechanism but does not immediately update the dynamic reference zero. The second recovery peak is used to confirm the automatic recovery of the baseline sequence rolling update mechanism after the respiratory rhythm stabilizes. The five most recent inspiratory start baselines are re-acquired in the following manner, and the median is calculated:

[0083] The accumulated zero-point drift during the freezing period is corrected to the correct baseline level in one go and used for multi-timescale joint judgment of respiratory abnormalities.

[0084] For example: no effective breathing for 20 consecutive seconds indicates a baseline sequence Stop recording new troughs, median The lock remains unchanged; however, long-term sensor data acquisition will still slowly generate zero-point drift, and real-time pressure data inherently contains offset errors. Since the baseline value is locked, the error continues to accumulate. At this point, correction is needed: for the first valid peak, only enable the baseline sequence. Data reception, median not updated The baseline sequence is not modified for the time being; the second valid peak: the system stores the new valid trough in the baseline sequence. The system automatically removes the oldest historical data and recalculates the median using the updated five troughs to generate a completely new median. The newly calculated median directly replaces the old freeze baseline for the freeze period, using the latest compliance baseline to smooth out the zero-point offset accumulated during the freeze period in one go.

[0085] When the invention detects consecutive invalid respiratory peaks exceeding a preset threshold (e.g., 20 seconds), it automatically freezes the dynamic baseline update to maintain the dynamic reference zero point. To prevent zero-point drift caused by accumulated sensor noise, a gradual strategy is adopted after breathing resumes: "recording is only restarted during the first recovery peak, and rolling updates are only resumed during the second recovery peak," achieving a one-time correction of the baseline level. This mechanism ensures data stability during apnea and avoids jump errors during recovery.

[0086] This invention maintains both a short-term respiratory rate sliding window and a long-term respiratory rate sliding window. In this embodiment, the short-term respiratory rate sliding window is 15 seconds, and the long-term respiratory rate sliding window is 60 seconds. For each valid respiratory segment identified by the verification in step 3, the total number of valid respiratory peaks in the short-term and long-term respiratory rate sliding windows is counted. The total number of valid respiratory peaks in the short-term respiratory rate sliding window is divided by the duration of the short-term respiratory rate sliding window to obtain the instantaneous average respiratory rate in the short-term respiratory rate sliding window. The total number of valid respiratory peaks in the long-term respiratory rate sliding window is divided by the duration of the long-term respiratory rate sliding window to obtain the instantaneous average respiratory rate in the long-term respiratory rate sliding window. The instantaneous average respiratory rate in the short-term respiratory rate sliding window is recorded as the short-term average respiratory rate, and the instantaneous average respiratory rate in the long-term respiratory rate sliding window is recorded as the long-term average respiratory rate.

[0087] See Figure 5 The alarm triggering conditions set by this invention include bradyspnea alarm triggering conditions and apnea alarm triggering conditions. When both the short-term average respiratory rate and the long-term average respiratory rate are lower than a certain value per minute... Once, and lasting longer than A short window period is required to meet the bradyventricular breathing alarm triggering conditions. In this embodiment... Indicates the number of breaths. Indicates the number of short-time window periods. , If the duration of no valid peak marking exceeds the no-breathing alarm threshold (in this embodiment, the no-breathing alarm threshold is 20 seconds), and during this 20-second period, waveform amplitude analysis confirms that no candidate breathing segments meet the amplitude validity judgment condition of step 2, it indicates that the breathing apnea alarm trigger condition is met.

[0088] When any alarm condition is met for the first time, the system does not immediately output an alarm, but enters a waiting confirmation state, continuously monitoring for a subsequent complete short-time window period. Only when the monitoring results for each second within this confirmation period meet the same alarm condition will a respiratory abnormality alarm signal be finally output and recorded in the log. At the same time, the original waveform segments 10 seconds before and after the alarm time are saved to the database for verification.

[0089] If, during the confirmation period, any normal respiratory peak that passes morphological verification occurs, or if there is a significant fluctuation in the amplitude of the candidate segment, the current pending confirmation status is revoked, thereby significantly reducing the false alarm rate caused by accidental interference or brief breath-holding.

[0090] Step 4 simultaneously maintains short-term (15-second) and long-term (60-second) sliding windows for respiratory rate, calculates the instantaneous average respiratory rate for each, and sets two alarm conditions: bradyventricular contraction and apnea. An alarm confirmation mechanism is introduced—continuous monitoring is performed within the first short-term window period, and an alarm is only output if the conditions are met throughout the entire period; otherwise, the pending trigger state is revoked. This mechanism significantly reduces the false alarm rate caused by accidental interference or brief breath-holding.

[0091] Example 2 See Figure 6 This embodiment provides a respiratory pressure measurement system, including: The waveform preprocessing module is used to process the respiratory pressure waveform to obtain candidate respiratory segments that conform to the respiratory rhythm.

[0092] The process of processing respiratory pressure waveforms to obtain candidate respiratory segments that conform to the respiratory rhythm includes: According to the normal human respiratory rate range, the potential respiratory waveform components are separated from the respiratory pressure waveform, and the first-order difference of the potential respiratory waveform components is calculated to obtain the first-order derivative sequence. Based on the derivative zero-crossing detection rule, the rising and falling inflection points of the respiratory pressure waveform are initially located. Based on the physiological constraints of a single human respiratory cycle, the time interval between adjacent rising inflection points is screened, and time segments within the physiological constraints of a single human respiratory cycle are marked as candidate respiratory segments that conform to the respiratory rhythm.

[0093] The normal human respiratory rate ranges from 0.1Hz to 0.8Hz; the physiological constraint of a single human respiratory cycle is 1.2 seconds to 5 seconds; time segments with a time interval of less than 1.2 seconds between adjacent rising edge inflection points are judged as high-frequency interference and discarded; time segments between 1.2 seconds and 5 seconds are marked as candidate respiratory segments that conform to the respiratory rhythm; and time segments with a time interval of more than 5 seconds between adjacent rising edge inflection points are temporarily stored in the observation buffer and a no-breathing timer is started.

[0094] The adaptive amplitude verification module is used to calculate the amplitude of each candidate respiratory segment, establish an amplitude sliding history queue based on historical amplitudes, calculate the dynamic average amplitude of the amplitude sliding history queue, and judge the amplitude validity of the current candidate respiratory segment based on the dynamic average amplitude to obtain a qualified amplitude candidate segment.

[0095] The amplitude of each candidate respiratory segment is calculated as follows: calculate the global maximum and global minimum values ​​of the respiratory pressure waveform within each candidate respiratory segment, and subtract the global minimum value from the global maximum value to obtain the amplitude of each candidate respiratory segment.

[0096] The method for determining the validity of current candidate respiratory segments based on dynamic average amplitude is as follows: Candidate respiratory segments whose amplitudes meet the validity judgment criteria are identified as qualified candidate segments. The validity judgment criteria are: , For amplitude, This is the dynamic average amplitude.

[0097] The waveform morphology matching verification module is used to normalize the amplitude of the actual respiratory rising segment waveform in the amplitude qualified candidate segments, calculate the normalized cross-correlation coefficient between the normalized rising edge waveform sequence and the standard half-sine wave template, and determine the amplitude qualified candidate segments with a normalized cross-correlation coefficient greater than or equal to the morphology matching threshold as valid respiratory segments.

[0098] The method for normalizing the amplitude of the actual respiratory rise segment waveform in the candidate segments with qualified amplitude is as follows:

[0099] in, This is the rising edge waveform sequence after amplitude normalization. The rising edge of the candidate segment with acceptable amplitude is the original pressure sampling sequence. , This represents the total number of sampling points during the ascending respiratory rate. , These are the minimum and maximum pressure values ​​within the actual respiratory rise waveform of the candidate segments with acceptable amplitude, respectively.

[0100] The method for calculating the normalized cross-correlation coefficient between the amplitude-normalized rising edge waveform sequence and the standard half-sine wave template is as follows:

[0101] in, To normalize the cross-correlation coefficient, This is the rising edge waveform sequence after amplitude normalization. The rising edge waveform sequence after amplitude normalization The overall average, These are sampled values ​​of a standard half-sine wave template waveform. Sampled values ​​of standard half-sine wave template waveform The overall average, , This represents the total number of sampling points during the ascending respiratory tract.

[0102] The standard half-sine wave template is:

[0103] in, This is a standard half-sine wave template waveform. For the real-time sampling time variable during the fragment's ascent process, , This represents the total number of sampling points during the ascending respiratory rate. It is the circumference constant. .

[0104] The morphological matching threshold is 0.85. After determining the qualified candidate segments with normalized cross-correlation coefficients greater than or equal to the morphological matching threshold as valid respiratory segments, the amplitude of the valid respiratory segments is confirmed as valid amplitude, and the valid amplitude is pushed into the amplitude sliding history queue. The amplitude sliding history queue automatically pops the earliest stored data to maintain the constant length of the amplitude sliding history queue, and at the same time outputs the valid peak time and peak pressure value of the valid respiratory segment.

[0105] The dynamic baseline calibration module is used to continuously record the trough pressure values ​​of multiple effective respiratory segments and form a baseline sequence. The median in the baseline sequence is used as the dynamic reference zero point. The respiratory pressure display value is obtained by subtracting the dynamic reference zero point from the collected real-time pressure data.

[0106] The median in the baseline sequence is calculated as follows: Sort the trough pressure values ​​of an odd number of effective breathing segments in ascending order:

[0107] , ... ... This represents the trough pressure values ​​of the odd-numbered effective respiratory segments in the baseline sequence. It is a positive integer greater than or equal to 1; the trough pressure value As the median in the baseline sequence.

[0108] During the continuous recording of trough pressure values ​​of multiple valid respiratory segments, invalid respiratory peaks are marked. When the time for which an invalid respiratory peak is marked exceeds the no-breathing alarm threshold, the baseline sequence update is stopped and the dynamic reference zero is frozen, keeping it at the dynamic reference zero calculated from the most recent valid respiratory cycle. When two consecutive valid respiratory peaks are detected, the first valid respiratory peak is used to restart the baseline recording mechanism but does not immediately update the dynamic reference zero. The second valid respiratory peak is used to confirm that the respiratory rhythm is stable and then automatically resume the baseline sequence rolling update mechanism, re-acquire the most recent inspiratory start baseline, and calculate the median.

[0109] The dual-window respiratory rate abnormality monitoring module is used to calculate the instantaneous average respiratory rate of the effective respiratory segment within the short-term respiratory rate sliding window and the long-term respiratory rate sliding window, and output a respiratory abnormality alarm signal in combination with the alarm triggering conditions.

[0110] Alarm triggering conditions include bradyventricular breathing and apnea. A bradyventricular breathing alarm is triggered when both the short-term and long-term average respiratory rates are below 8 breaths per minute for more than two short-term window periods. An apnea alarm is triggered when there are no valid peak markers for more than 20 seconds, and waveform amplitude analysis confirms that no candidate respiratory segments meeting the amplitude validity criteria appear during this 20-second period.

[0111] When any alarm condition is met for the first time, the system does not immediately output an alarm, but enters a waiting confirmation state, continuously monitoring for a subsequent complete short-time window period. Only when the monitoring results for each second within this confirmation period meet the same alarm condition will a respiratory abnormality alarm signal be finally output and recorded in the log. At the same time, the original waveform segments 10 seconds before and after the alarm time are saved to the database for verification.

[0112] If, during the confirmation period, any normal respiratory peak that passes morphological verification occurs, or if there is a significant fluctuation in the amplitude of the candidate segment, the current alarm pending state (waiting for confirmation state) will be cancelled, thereby significantly reducing the false alarm rate caused by accidental interference or brief breath-holding.

[0113] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for measuring respiratory pressure, characterized in that: include: The respiratory pressure waveform is processed to obtain candidate respiratory segments that conform to the respiratory rhythm; The amplitude of each candidate respiratory segment is calculated, and a sliding historical amplitude queue is established based on the historical amplitude. The dynamic average amplitude of the sliding historical amplitude queue is calculated, and the validity of the current candidate respiratory segment is judged based on the dynamic average amplitude to obtain the candidate segment with qualified amplitude. The waveform of the actual respiratory rising segment in the amplitude qualified candidate segment is normalized. The normalized cross-correlation coefficient between the normalized rising edge waveform sequence and the standard half-sine wave template is calculated. The amplitude qualified candidate segment with the normalized cross-correlation coefficient greater than or equal to the morphological matching threshold is determined as a valid respiratory segment. The trough pressure values ​​of multiple valid respiratory segments are continuously recorded to form a baseline sequence. When the time without valid respiratory peaks exceeds the no-breathing alarm threshold, the baseline sequence update is stopped and the dynamic reference zero point is frozen, keeping it at the dynamic reference zero point calculated from the most recent valid respiratory cycle. When two consecutive valid respiratory peaks are detected, the first valid respiratory peak is used to restart the baseline recording mechanism but does not immediately update the dynamic reference zero point. The second valid respiratory peak is used to confirm that the respiratory rhythm is stable, and the baseline sequence rolling update mechanism is automatically restored. The most recent inspiratory start baseline is re-acquired, and the median is calculated. The median in the baseline sequence is used as the dynamic reference zero point. The respiratory pressure display value is obtained by subtracting the dynamic reference zero point from the acquired real-time pressure data. It calculates the instantaneous average respiratory rate of the effective respiratory segment within the short-term and long-term respiratory rate sliding windows, and outputs a respiratory abnormality alarm signal based on the alarm triggering conditions.

2. The respiratory pressure measurement method according to claim 1, characterized in that: The process of processing respiratory pressure waveforms to obtain candidate respiratory segments that conform to the respiratory rhythm includes: According to the normal human respiratory rate range, the potential respiratory waveform components are separated from the respiratory pressure waveform, and the first-order difference of the potential respiratory waveform components is calculated to obtain the first-order derivative sequence. Based on the derivative zero-crossing detection rule, the rising and falling inflection points of the respiratory pressure waveform are initially located. Based on the physiological constraints of a single human respiratory cycle, the time interval between adjacent rising inflection points is screened, and time segments within the physiological constraints of a single human respiratory cycle are marked as candidate respiratory segments that conform to the respiratory rhythm.

3. The respiratory pressure measurement method according to claim 2, characterized in that: The normal human respiratory rate ranges from 0.1Hz to 0.8Hz; the physiological constraint of a single human respiratory cycle is 1.2 seconds to 5 seconds; time segments with a time interval of less than 1.2 seconds between adjacent rising edge inflection points are judged as high-frequency interference and discarded; time segments between 1.2 seconds and 5 seconds are marked as candidate respiratory segments that conform to the respiratory rhythm; and time segments with a time interval of more than 5 seconds between adjacent rising edge inflection points are temporarily stored in the observation buffer and a no-breathing timer is started.

4. The respiratory pressure measurement method according to claim 1, characterized in that: The amplitude of each candidate respiratory segment is calculated as follows: calculate the global maximum and global minimum values ​​of the respiratory pressure waveform within each candidate respiratory segment, and subtract the global minimum value from the global maximum value to obtain the amplitude of each candidate respiratory segment.

5. The respiratory pressure measurement method according to claim 1, characterized in that: The method for determining the validity of current candidate respiratory segments based on dynamic average amplitude is as follows: Candidate respiratory segments whose amplitudes meet the validity judgment criteria are identified as qualified candidate segments. The validity judgment criteria are: , For amplitude, This is the dynamic average amplitude.

6. The respiratory pressure measurement method according to claim 1, characterized in that: The method for normalizing the amplitude of the actual respiratory rise segment waveform in the candidate segments with qualified amplitude is as follows: in, This is the rising edge waveform sequence after amplitude normalization. The rising edge of the candidate segment with acceptable amplitude is the original pressure sampling sequence. , This represents the total number of sampling points during the ascending respiratory rate. , These are the minimum and maximum pressure values ​​within the actual respiratory rise waveform of the candidate segments with acceptable amplitude, respectively.

7. The respiratory pressure measurement method according to claim 1, characterized in that: The method for calculating the normalized cross-correlation coefficient between the amplitude-normalized rising edge waveform sequence and the standard half-sine wave template is as follows: in, To normalize the cross-correlation coefficient, This is the rising edge waveform sequence after amplitude normalization. The rising edge waveform sequence after amplitude normalization The overall average, These are sampled values ​​of a standard half-sine wave template waveform. Sampled values ​​of standard half-sine wave template waveform The overall average, , This represents the total number of sampling points during the ascending respiratory rate. The standard half-sine wave template is: in, This is a standard half-sine wave template waveform. For the real-time sampling time variable during the fragment's ascent process, , This represents the total number of sampling points during the ascending respiratory rate. It is the circumference constant. .

8. The respiratory pressure measurement method according to claim 1, characterized in that: The morphological matching threshold is 0.

85. After determining the qualified candidate segments with normalized cross-correlation coefficients greater than or equal to the morphological matching threshold as valid respiratory segments, the amplitude of the valid respiratory segments is confirmed as valid amplitude, and the valid amplitude is pushed into the amplitude sliding history queue. The amplitude sliding history queue automatically pops the earliest stored data to maintain the constant length of the amplitude sliding history queue, and at the same time outputs the valid peak time and peak pressure value of the valid respiratory segment.

9. The respiratory pressure measurement method according to claim 1, characterized in that: The median in the baseline sequence is calculated as follows: Sort the trough pressure values ​​of an odd number of effective breathing segments in ascending order: , ... ... This represents the trough pressure values ​​of the odd-numbered effective respiratory segments in the baseline sequence. It is a positive integer greater than or equal to 1; the trough pressure value As the median in the baseline sequence.

10. The respiratory pressure measurement method according to claim 1, characterized in that: Alarm triggering conditions include bradyventricular breathing alarm triggering conditions and apnea alarm triggering conditions. The instantaneous average respiratory rate of the effective respiratory segment within the short-term respiratory rate sliding window and the long-term respiratory rate sliding window are calculated respectively to obtain the short-term average respiratory rate and the long-term average respiratory rate. When both the short-term average respiratory rate and the long-term average respiratory rate are lower than [a certain value] per minute... Once, and lasting longer than A short window period is required to meet the conditions for triggering the bradybreathing alarm. Indicates the number of breaths. Indicates the number of short-time window periods; When the duration of no valid peak markers exceeds the no-breathing alarm threshold, and there are no candidate breathing segments that meet the amplitude validity judgment conditions during the no-breathing alarm threshold period, the apnea alarm trigger condition is met.

11. A respiratory pressure measurement system, characterized in that: include: The waveform preprocessing module is used to process the respiratory pressure waveform to obtain candidate respiratory segments that conform to the respiratory rhythm; The adaptive amplitude verification module is used to calculate the amplitude of each candidate respiratory segment, establish an amplitude sliding history queue based on historical amplitudes, calculate the dynamic average amplitude of the amplitude sliding history queue, and judge the amplitude validity of the current candidate respiratory segment based on the dynamic average amplitude to obtain the candidate segment with qualified amplitude. The waveform morphology matching verification module is used to normalize the amplitude of the actual respiratory rising segment waveform in the amplitude qualified candidate segment, calculate the normalized cross-correlation coefficient between the normalized rising edge waveform sequence and the standard half-sine wave template, and determine the amplitude qualified candidate segment with the normalized cross-correlation coefficient greater than or equal to the morphology matching threshold as a valid respiratory segment. The dynamic baseline calibration module continuously records the trough pressure values ​​of multiple effective respiratory segments to form a baseline sequence. When the time without effective respiratory peaks exceeds the no-breathing alarm threshold, the baseline sequence update is stopped and the dynamic reference zero point is frozen, keeping it at the dynamic reference zero point calculated from the most recent effective respiratory cycle. When two consecutive effective respiratory peaks are detected, the first effective respiratory peak is used to restart the baseline recording mechanism but does not immediately update the dynamic reference zero point. The second effective respiratory peak is used to confirm that the respiratory rhythm is stable, and the baseline sequence rolling update mechanism is automatically restored, the most recent inspiratory start baseline is re-acquired, and the median is calculated. The median in the baseline sequence is used as the dynamic reference zero point, and the respiratory pressure display value is obtained by subtracting the dynamic reference zero point from the acquired real-time pressure data. The dual-window respiratory rate abnormality monitoring module is used to calculate the instantaneous average respiratory rate of the effective respiratory segment within the short-term respiratory rate sliding window and the long-term respiratory rate sliding window, and output a respiratory abnormality alarm signal in combination with the alarm triggering conditions.

Images