A time-domain-based method for detecting respiratory events in adults

Abstract

The present invention relates to a method for detecting adult respiratory events based on time domain. The problem in the prior art that manual identification of respiratory events takes a long time, has inconsistent standards, and has low accuracy is solved. The method includes obtaining oral and nasal airflow signals, chest movement signals, and blood oxygen saturation signals; performing time domain division analysis on the oral and nasal airflow signals, obtaining signal amplitude data, and recording signal events; preliminarily classifying the oral and nasal airflow signal events according to the oral and nasal airflow signals and airflow snoring signals, and classifying them into temporary apnea events and temporary hypopnea events; combining temporary hypopnea events with blood oxygen saturation information to determine hypopnea events; comparing temporary apnea events with chest movement signal amplitude data to determine the type of apnea events. Compared with manual identification of respiratory time, the present invention has a shorter identification time, a unified identification standard, and a more accurate identification structure, so that a structure can be quickly and accurately automatically analyzed for reference by doctors.

Description

Technical Field

[0001] The present invention relates to the technical field of physiological parameter analysis, and in particular to a time domain-based adult respiratory event detection method. Background Art

[0002] As people pay more and more attention to sleep, sleep medicine has become a new hot spot in clinical medicine. Objective detection of sleep is the basic condition for sleep medicine and sleep-related research. Polysomnography (PSG) is the internationally recognized gold standard for diagnosing sleep apnea hypopnea syndrome. The identification of nocturnal sleep respiratory events is an important factor in disease diagnosis.

[0003] Respiratory events are divided into hypopnea and apnea, and apnea is further divided into central apnea, obstructive apnea, and mixed apnea. The criteria for distinguishing each respiratory event in AASM are as follows:

[0004] Hypopnea: Oral and nasal airflow decreases by ≥30% compared with baseline, lasting ≥10 seconds, and is accompanied by a decrease in blood oxygen saturation of ≥3%.

[0005] Apnea: Oral and nasal airflow decreases by ≥90% compared with baseline and lasts for ≥10s.

[0006] Central apnea: Apnea occurs when there is no effort to breathe.

[0007] Obstructive apnea: A period of apnea during which there is a continuous effort to breathe.

[0008] Mixed apnea: During apnea, there is no effort to breathe at the beginning, but then there is a continuous effort to breathe.

[0009] In reality, respiratory events are usually identified manually, but there are problems such as long identification time, inconsistent identification standards, and low structural accuracy of the identification. Summary of the invention

[0010] The present invention mainly solves the problems in the prior art that manual identification of respiratory events takes a long time, has inconsistent standards, and has low accuracy, and provides an adult respiratory event detection method based on time domain.

[0011] The above technical problem of the present invention is mainly solved by the following technical solution: a method for detecting adult respiratory events based on time domain, comprising the following steps:

[0012] S1. Obtain oral and nasal airflow signals, chest movement signals and blood oxygen saturation signals;

[0013] S2. Perform time domain analysis on the oral and nasal airflow signals, obtain signal amplitude data and record signal events;

[0014] S3. Preliminarily classify the oronasal airflow signal events according to the oronasal airflow signal and the airflow snoring signal into temporary apnea events and temporary hypopnea events;

[0015] S4. Determine a hypopnea event by combining the provisional hypopnea event with the blood oxygen saturation information;

[0016] The type of apnea event is determined by comparing the temporary apnea event with the chest movement signal amplitude data. The present invention determines the specific types of hypopnea events and apnea events according to the amplitude of the segments by performing time domain division analysis on the oral and nasal airflow signals, chest movement signals and blood oxygen saturation signals. Compared with manual determination of breathing time, the present invention has a shorter determination time, a unified determination standard and a more accurate determination structure, so that a structure can be quickly and accurately automatically analyzed for reference by doctors.

[0017] As a preferred solution, the time domain division analysis process in step S2 includes:

[0018] S21. Divide the signal into n segments according to one breathing time, and obtain the signal segment sequence W(w n ), w n In this scheme, the preferred breathing time is 2.5 seconds. A normal adult breathes between 16 and 20 times per minute, that is, one breath every 3 to 4 seconds. Using a 2.5-second window, which is slightly shorter than the time of one breath, can more accurately estimate the change in amplitude.

[0019] S22. The difference between the maximum and minimum values ​​in each segment is taken as the segment amplitude, and the segment amplitude sequence A (a n ), a n is the segment amplitude; the segment amplitude sequence can reflect the amplitude change of the signal every 2.5 seconds.

[0020] S23. Set the amplitude baseline a base , with amplitude sequence A(a n ) as the initial value of the amplitude baseline, and traverse the segment amplitude sequence A(a n );

[0021] If the current amplitude a k >Amplitude baseline a base , then get the two amplitudes a adjacent to the current amplitude k-1 、a k+1 , the amplitude baseline is updated to a base =(a k-1 +ak +a k+1 ) / 3; The amplitude baseline represents the normal value of the amplitude and is used for comparison with the amplitude decrease.

[0022] If the current amplitude a k <μa base , where μ is the set proportional coefficient, 0≤μ≤1, and starts recording signal events. If the amplitude a lasts for m segments k 、a k-1 …a k+m All less than μa base , until a k+m+1 >μa base , signal event recording ends, and the amplitude baseline is updated to a base = a k+m+1 In the time domain division analysis of the oral and nasal airflow signals, μ takes a value of 0.7, and in the time domain division analysis of the chest motion signals, μ takes a value in the range of 0.1-0.2, preferably 0.2.

[0023] As a preferred solution, the oral and nasal airflow signals are divided and analyzed in the time domain to record respiratory events, and the process includes:

[0024] Record possible respiratory events k , w k+1 ,…,w k+m}, since the segment length is 2.5 seconds, the event range may be inaccurate. By calculating the w of two adjacent segments k-1 and w k+m+1 The event range is adjusted by the differential of , extending the range to the left and right sides, and the extended segments are recorded as w left and w right , whose lengths are Δt left and Δt right ;

[0025] Judgment (2.5*m+Δt left +Δt right ) is greater than 10s, if so, then m+2 segments { w left , w k , w k+1 ,…,w k+m , w right} is recorded as a respiratory event, otherwise, the event is not recorded.

[0026] As a preferred solution, the specific process of step S3 includes:

[0027] S31. For respiratory events { w left , w k , w k+1 ,…,wk+m , w right}, the detection amplitude is less than 0.7a base And greater than 0.1a base The number of these fragments is recorded as n1, and the amplitude is tested to see if it is less than or equal to 0.1a base The number of these fragments is recorded as n2;

[0028] S32. Determine whether n2≥n1, if so, respiratory event {w left , w k , w k+1 ,…,w k+m , w right} is recorded as the first apnea event ap1, if not, the respiratory event { w left , w k , w k+1 ,…,w k+m , w right} recorded as the first hypopnea event hp1;

[0029] S33. Perform a 5 Hz high-pass filter on the oral and nasal airflow signal to obtain an airflow snoring signal, combine the first apnea event ap1 with the airflow snoring signal, wherein the event in the first apnea event ap1 that includes the airflow snoring signal is recorded as the second hypopnea event hp2, and the remaining first apnea event ap1 is recorded as the second apnea event ap2;

[0030] S34. Record the first hypopnea event hp1 and the second hypopnea event hp2 as provisional hypopnea events hp, and record the second apnea event ap2 as provisional apnea event ap.

[0031] As a preferred solution, the process of determining a hypopnea event in step S4 includes:

[0032] Detect whether the blood oxygen saturation decreases by Y within X seconds after the start of each tentative hypopnea event. If so, record it as a hypopnea event. If not, do not record it. Get all hypopnea events, where X is 45 seconds and Y is 3%.

[0033] As a preferred solution, the process of determining the type of apnea event in step S4 includes:

[0034] S41. performing time domain division analysis on the chest motion signal, obtaining a segment amplitude sequence and an amplitude baseline of the chest motion signal, comparing the amplitude of each segment in the segment amplitude sequence with the amplitude baseline, and marking segments in which the amplitude decreases by more than 90%;

[0035] S42. Compare apnea events with chest movement signals,

[0036] If the time of the chest movement signal in the first half of the apnea event with an amplitude drop of more than 90% accounts for more than 60%, and the time of the chest movement signal in the second half of the apnea event with a normal amplitude accounts for more than 70%, then the apnea time is recorded as a mixed apnea event;

[0037] If the proportion of the time segment in which the amplitude of the chest motion signal drops by more than 90% within the apnea event range exceeds 70%, the apnea time is recorded as a central apnea event;

[0038] The apnea time that was not in the above two situations was recorded as obstructive apnea event.

[0039] Therefore, the advantages of the present invention are: according to the oral and nasal airflow signals, chest movement signals and blood oxygen saturation signals, by performing time domain division analysis on the oral and nasal airflow signals and chest movement signals, the specific types of hypopnea events and apnea events are judged according to the amplitude of the segments. Compared with manual judgment of breathing time, the present invention has a shorter judgment time, a unified judgment standard, and a more accurate judgment structure, so that a structure can be quickly and accurately analyzed automatically for the doctor's reference. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of a process of the present invention. DETAILED DESCRIPTION

[0041] The technical solution of the present invention is further specifically described below through embodiments and in conjunction with the accompanying drawings.

[0042] Example:

[0043] This embodiment is a method for detecting adult respiratory events based on time domain, such as Figure 1 As shown, the following steps are included:

[0044] S1. Obtain oral and nasal airflow signals, chest movement signals and blood oxygen saturation signals; perform 0.5 Hz low-pass filtering on the nose picking airflow signals and chest movement signals respectively.

[0045] S2. Perform time domain analysis on the oral and nasal airflow signals, obtain signal amplitude data and record signal events; the time domain analysis process includes:

[0046] S21. Divide the signal into n segments according to one breathing time, and obtain the signal segment sequence W(w n ), w n For fragments;

[0047] S22. The difference between the maximum and minimum values ​​in each segment is taken as the segment amplitude, and the segment amplitude sequence A (a n ), a nis the segment amplitude;

[0048] S23. Set the amplitude baseline a base , with amplitude sequence A(a n ) as the initial value of the amplitude baseline, and traverse the segment amplitude sequence A(a n );

[0049] If the current amplitude a k >Amplitude baseline a base , then get the two amplitudes a adjacent to the current amplitude k-1 、a k+1 , the amplitude baseline is updated to a base =(a k-1 +a k +a k+1 ) / 3;

[0050] If the current amplitude a k <μa base , where μ is the set proportional coefficient, 0≤μ≤1, and the time domain analysis of the oral and nasal airflow signal is performed, and μ takes a value of 0.7. Start recording signal events. If the amplitude a of m segments lasts k 、a k-1 …a k+m All less than μa base , until a k+m+1 >0.7a base , signal event recording ends, and the amplitude baseline is updated to a base = a k+m+1 .

[0051] The preferred breathing time is 2.5 seconds. A normal adult breathes between 16 and 20 times per minute, that is, one breath every 3 to 4 seconds. Using a 2.5 second window, which is slightly shorter than the time of one breath, can more accurately estimate the change in amplitude. The recorded signal event may constitute a breathing event, which is recorded as a possible breathing event {w k , w k+1 ,…,w k+m}. Since the segment length is 2.5 seconds, the event range may be inaccurate. By calculating the w k-1 and w k+m+1 The event range is adjusted by the differential of , extending the range to the left and right sides, and the extended segments are recorded as w left and w right , whose lengths are Δt left and Δt right ;

[0052] Judgment (2.5*m+Δt left +Δt right) is greater than 10s, if so, then m+2 segments { w left , w k , w k+1 ,…,w k+m , w right} is recorded as a respiratory event, otherwise, the event is not recorded.

[0053] S3. Preliminarily classify the oronasal airflow signal events according to the oronasal airflow signal and the airflow snoring signal into temporary apnea events and temporary hypopnea events. The specific process includes:

[0054] S31. For respiratory events { w left , w k , w k+1 ,…,w k+m , w right}, the detection amplitude is less than 0.7a base And greater than 0.1a base The number of these fragments is recorded as n1, and the amplitude is tested to see if it is less than or equal to 0.1a base The number of these fragments is recorded as n2;

[0055] S32. Determine whether n2≥n1, if so, respiratory event {w left , w k , w k+1 ,…,w k+m , w right} is recorded as the first apnea event ap1, if not, the respiratory event { w left , w k , w k+1 ,…,w k+m , w right} recorded as the first hypopnea event hp1;

[0056] S33. Perform a 5 Hz high-pass filter on the oral and nasal airflow signal to obtain an airflow snoring signal, combine the first apnea event ap1 with the airflow snoring signal, wherein the event in the first apnea event ap1 that includes the airflow snoring signal is recorded as the second hypopnea event hp2, and the remaining first apnea event ap1 is recorded as the second apnea event ap2;

[0057] S34. Record the first hypopnea event hp1 and the second hypopnea event hp2 as provisional hypopnea events hp, and record the second apnea event ap2 as provisional apnea event ap.

[0058] S4. Combine the tentative hypopnea event with the blood oxygen saturation information to determine the hypopnea event; the process includes:

[0059] Detect whether the blood oxygen saturation decreases by 3% within 45 seconds after the start of each temporary hypopnea event. If so, record it as a hypopnea event. If not, do not record it. All hypopnea events are obtained.

[0060] The type of apnea event is determined by comparing the temporary apnea event with the chest movement signal amplitude data. The process includes:

[0061] S41. Perform time domain division analysis on the chest motion signal according to steps S21-S23, where μ takes a value of 0.2. Obtain the segment amplitude sequence and amplitude baseline of the chest motion signal, compare the amplitude of each segment in the segment amplitude sequence with the amplitude baseline, and mark the segments where the amplitude drops by more than 90%;

[0062] S42. Compare apnea events with chest movement signals,

[0063] If the time of the chest movement signal in the first half of the apnea event with an amplitude drop of more than 90% accounts for more than 60%, and the time of the chest movement signal in the second half of the apnea event with a normal amplitude accounts for more than 70%, then the apnea time is recorded as a mixed apnea event;

[0064] If the time of the segment in which the amplitude of the chest movement signal drops by more than 90% within the apnea event range exceeds 70%, the apnea time is recorded as a central apnea event;

[0065] The apnea time that was not in the above two situations was recorded as obstructive apnea event.

[0066] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. A time-domain based adult respiratory event detection method, characterized in that: The following steps are included: S1. Obtain oral and nasal airflow signals, chest movement signals and blood oxygen saturation signals; S2. Perform time domain analysis on the oral and nasal airflow signals, obtain signal amplitude data and record signal events; including: S21. Divide the signal into n segments according to one breathing time, and obtain the signal segment sequence W(w n ); S22. Obtain the segment amplitude sequence A(a n ); S23. With amplitude sequence A(a n ) sets the initial value of the amplitude baseline, and traverses the segment amplitude sequence A(a n ); If the current amplitude a k >Amplitude baseline a base , then get the two amplitudes a adjacent to the current amplitude k-1 、a k+1 , the amplitude baseline is updated to a base =(a k-1 +a k +a k+1 ) / 3; If the current amplitude a k <μa base , μ is the set proportional coefficient, 0≤μ≤1, start recording signal events, if the amplitude a lasts for m+1 segments k 、a k+1 …a k+m All less than μa base , until a k+m+1 >μa base , signal event recording ends, and the amplitude baseline is updated to a base = a k+m+1 ; S3. Preliminary classification of oronasal airflow signal events is performed according to the oronasal airflow signal and the airflow snoring signal, and the events are classified into temporary apnea events and temporary hypopnea events.

2. The time-domain based adult respiratory event detection method according to claim 1, characterized in that The time domain division analysis process in step S2 includes: The difference between the maximum and minimum values ​​in each segment is taken as the segment amplitude.

3. The time-domain based adult respiratory event detection method according to claim 2, characterized in that In the time domain analysis of oral and nasal airflow signals, respiratory events are recorded. The process includes: Record possible respiratory events k , w k+1 ,…,w k+m }, calculate two adjacent fragments w k-1 and w k+m+1 , denoted as w left and w right , whose lengths are Δt left and Δt right ; Judgment (2.5*m+Δt left +Δt right ) is greater than 10s, if so, then m+3 segments { w left , w k , w k+1 ,…,w k+m , w right } is recorded as a respiratory event, otherwise, the event is not recorded.

4. The time-domain based adult respiratory event detection method according to claim 3, characterized in that The specific process of step S3 includes: S31. For respiratory events { w left , w k , w k+1 ,…,w k+m , w right }, the detection amplitude is less than 0.7a base And greater than 0.1a base The number of these fragments is recorded as n1, and the amplitude is tested to see if it is less than or equal to 0.1a base The number of these fragments is recorded as n2; S32. Determine whether n2≥n1, if so, respiratory event {w left , w k , w k+1 ,…,w k+m , w right } is recorded as the first apnea event ap1, if not, the respiratory event { w left , w k , w k+1 ,…,w k+m , w right } is recorded as the first hypopnea event hp1.

5. The time-domain based adult respiratory event detection method according to claim 4, characterized in that The specific process of step S3 includes: S33. Perform a 5 Hz high-pass filter on the oral and nasal airflow signal to obtain an airflow snoring signal, combine the first apnea event ap1 with the airflow snoring signal, wherein the event in the first apnea event ap1 that includes the airflow snoring signal is recorded as the second hypopnea event hp2, and the remaining first apnea event ap1 is recorded as the second apnea event ap2; S34. Record the first hypopnea event hp1 and the second hypopnea event hp2 as provisional hypopnea events hp, and record the second apnea event ap2 as provisional apnea event ap.

Images