CN117272171A

CN117272171A - Quantitative analysis method and device for 24-hour breathing mode

Info

Publication number: CN117272171A
Application number: CN202311219684.XA
Authority: CN
Inventors: 张政波; 王佳晨; 李丽轩; 李梦伟; 寇宇晴
Original assignee: Chinese PLA General Hospital
Current assignee: Chinese PLA General Hospital
Priority date: 2022-12-16
Filing date: 2023-09-21
Publication date: 2023-12-22

Abstract

The application provides a quantitative analysis method of a 24-hour breathing mode, which comprises the steps of obtaining a breathing signal of a subject through wearable equipment, and carrying out SQI quality evaluation on the breathing signal to obtain a breathing signal with high quality; acquiring parameters from the high quality respiratory signal; the breathing pattern of the subject is quantitatively reflected by the values of these parameters. The quantitative analysis method and the quantitative analysis device for the 24-hour breathing mode provide objective indexes for quantitative breathing and effective coordinates and scales for 24-hour breathing mode research.

Description

Quantitative analysis method and device for 24-hour breathing mode

Technical Field

The present application relates to analysis of 24-hour breathing patterns, and more particularly to a method and apparatus for quantitative analysis of 24-hour breathing patterns.

Background

Breathing pattern refers to the amplitude, periodicity, chest-abdomen contribution and coordination of breathing, which can be considered as a result of a combination of various systems, including central and peripheral chemical reactions, cardiac systems and respiratory systems. Monitoring the breathing pattern is critical for respiratory dysfunction and failure patients, as respiratory changes are often important indicators of disease progression or improvement. In diseased cases, an increase in respiratory rate may be associated with hypoxia, pain, cardiac insufficiency or metabolic disorders such as lactic acidosis. The reduced respiratory rate may be associated with depression of the nervous system or even more severe hypoxia. Short-time monitoring studies are currently common, while long-time (e.g., 24 hours) is an emerging monitoring modality. The 24-hour monitoring can be used as early warning for patient disease deterioration and improving follow-up consciousness. For example, weekly changes in respiratory rate are reported to increase before heart failure exacerbates. Monitoring the change in respiratory rate facilitates diagnosing high risk patients 24 hours earlier than the actual event. In addition, respiratory pattern monitoring is reported to be useful for detecting analgesic-induced respiratory depression within 24 hours after surgery.

There were few studies of 24-hour breathing patterns compared to numerous studies of ECG and HRV. On the one hand, the measurement method using the mask changes the natural respiration, and 24-hour respiration monitoring cannot be performed outside the laboratory. On the other hand, there is currently no acceptable breath signal analysis standard and range, as the frequency, amplitude and pause of breathing can be varied independently of each other. In recent years, it has become increasingly appreciated that respiratory monitoring is critical to patient health status assessment. Cesareo et al (A. Cesareo et al., "A wearabledevice for breathing frequency monitoring: A pilot study on patients with muscular dystrophy," Sensors, 2020) studied the feasibility of using inertial sensor-based devices to monitor respiratory function in muscular dystrophy patients and demonstrated a strong correlation between respiratory rate and respiratory function primary index, both in the clinic and at home. Angelucci et al (A. Angelucci et al., "A home telemedicine system for continuous respiratory monitoring," IEEE J-BHI, vol.25, no.4,2020) propose a home telemedicine respiratory monitoring system. Massaroni et al (c.massaroni et al., "Non-contact monitoring of breathing pattern and respiratory rate via RGB signal measurement," Sensors, 2019) propose a respiratory measurement system by analyzing video data acquired by an RGB camera built into a notebook computer. Mann 'ee et al (D.C.Mann' ee, F.de Jongh, and H.van Helvoort, "Telemonitoring techniques for lung volume measurement: accuracy, artifacts and effort," Frontiers in Digital Health, vol.2, p.16, 2020) review the recent progress of remote monitoring techniques for measuring lung capacity. However, clinical application studies of 24-hour breathing patterns are still in the start phase.

With the rapid development of mobile computing technology, the advent of wearable devices has made it possible to collect 24-hour respiratory signals outside of a laboratory. However, the potential value of how to mine 24-hour respiratory signals is still not well addressed.

Disclosure of Invention

In view of the above problems, the present application aims to provide a quantitative analysis method and a device for a 24-hour breathing pattern, which provide objective basis and scale for quantitative analysis of the breathing pattern.

The application provides a quantitative analysis method of a 24-hour breathing mode, which comprises the steps of obtaining a breathing signal of a subject through wearable equipment, and carrying out SQI quality evaluation on the breathing signal to obtain a breathing signal with high quality; the following parameters are obtained from the high quality respiratory signal:

BR_mean、BR_cv、BR_skewness、BR_kurtosis、

MV_mean、MV_cv、MV_skewness、MV_kurtosis、

AB_mean、AB_cv、AB_skewness、AB_kurtosis、

IE_mean、IE_cv、IE_skewness、IE_kurtosis、

LF、HF、LF_HF、nLF、nHF、TP、vLF、

SD1、SD2、SD2/SD1、

SampEn、Area_1_5、Area_6_20；

the breathing pattern of the subject is quantitatively reflected by the values of the above parameters.

The application also provides a quantitative analysis device of 24-hour breathing mode, which comprises a signal acquisition unit, an SQI quality evaluation unit, a parameter acquisition unit and a parameter comparison unit:

the signal acquisition unit acquires a respiratory signal of a subject from the wearable device;

the SQI quality evaluation unit performs SQI quality evaluation on the respiratory signal to obtain a respiratory signal with high quality;

the parameter acquisition unit acquires the following parameters from the high-quality respiratory signal:

BR_mean、BR_cv、BR_skewness、BR_kurtosis、

MV_mean、MV_cv、MV_skewness、MV_kurtosis、

AB_mean、AB_cv、AB_skewness、AB_kurtosis、

IE_mean、IE_cv、IE_skewness、IE_kurtosis、

LF、HF、LF_HF、nLF、nHF、TP、vLF、

SD1、SD2、SD2/SD1、

SampEn、Slope_1_5、Area_1_5、Area_6_20；

the parameter comparison unit quantitatively reflects the breathing pattern of the subject by the values of the above parameters.

Preferably, the respiration of the subject is divided into four states, which are a daytime state, a nighttime state, a standing upright state, a lying position state;

the reference ranges of the parameters are different in different states.

Preferably, in the daytime condition,

the reference range of BR_mean is 21.3+ -3.0;

the reference range of BR_cv is 22.1+ -3.5;

the reference range of BR_skew is 0.8+ -0.4;

the reference range of BR_kurtosis is 0.4[0.2,1.2];

the MV_mean reference range is 44700.8 +/-16377.4;

the reference range for MV_cv is 49.1[43.2,58.5];

the reference range of MV_Shewness is 1.3+ -0.6;

the reference range of MV_kurtosis is 2.1[0.7,4.2];

the reference range of AB_mean is 0.4+/-0.2;

the reference range for AB_cv is 39.5[32.9,46.6];

the reference range of AB_skew is 0.6[0.1,1.1];

the reference range of AB_kurtosis is-0.2 < -0.7,1.2 >;

the reference range of IE_mean is 0.7+ -0.1;

the reference range of IE_cv is 25.1[22.6,28.6];

the reference range of IE_skew is 1.2[0.9,1.7];

the reference range of IE_kurtosis is 3.2[2.1,6.1];

the reference range of LF is 56887.3[48891.5,67446.2]; the reference range for HF is 25005.8[20613.1,27915.5]; the reference range of LF_HF is 2.4[2.2,2.7];

nLF with a reference range of 70.4 + -3.3;

the reference range of nHF is 29.6+/-3.3;

the reference range of TP is 91280.9[79401.6,103554.1]; the reference range of vLF is 8745.7[7617.3,10473.3]; the reference range for SD1 is 536.9[480.0,592.7];

SD2 has a reference range of 969.9[865.9,1107.8];

the reference range of SD2/SD1 is 1.8[1.7,2.0];

the reference range for SampEn is 1.4[1.3,1.6];

the reference range of area_1_5 is 6.6[6.1,7.1]; the reference range of area_6_20 is 16.7[15.3,17.5]; in the night-time state,

the reference range of BR_mean is 16.3[15.1,17.8];

the reference range of BR_cv is 19.8+ -5.2;

the reference range of BR_skew is 1.5+ -0.6; the reference range of BR_kurtosis is 3.3[1.9,6.6];

the MV_mean reference range is 22514.3 +/-8087.3;

the reference range for MV_cv is 64.2[53.3,76.5];

the reference range of MV_Shewness is 2.4[1.8,3.2];

the reference range of MV_kurtosis is 9.5[4.3,15.0];

the reference range of AB_mean is 0.6[0.5,0.7]; the reference range for AB_cv is 23.2[18.2,29.0];

the reference range of AB_skew is-0.5 [ -1.1, -0.0];

the reference range of AB_kurtosis is 0.5 < -0.3,1.8 >;

the reference range of IE_mean is 0.8[0.7,0.8]; the reference range of IE_cv is 28.1[24.2,32.2]; the reference range of IE_skew is 1.4[1.0,2.0];

the reference range of IE_kurtosis is 4.6[1.7,8.2];

the reference range of LF is 45018.5[36540.3,76241.9]; the reference range for HF is 14733.1[11784.6,22443.2]; the reference range of LF_HF is 3.4+/-0.6;

nLF has a reference range of 76.9[74.1,78.8];

the reference range of nHF is 23.1[21.2,25.9];

the reference range for TP is 65852.8[53659.4,109706.9]; the reference range for vLF is 6944.0[5619.5,10445.9]; the reference range for SD1 is 548.5[459.8,760.2];

the reference range for SD2 is 1067.6[885.3,1344.3];

the reference range of SD2/SD1 is 1.8+/-0.4;

the reference range for SampEn is 1.4[1.2,1.5];

the reference range of area_1_5 is 5.7[4.9,6.5]; the reference range of area_6_20 is 11.4[9.4,12.8];

in the state of resting and standing upright,

the reference range of BR_mean is 19.8+ -2.7;

the reference range for BR_cv is 14.8[13.2,16.5];

the reference range of BR_skew is 0.3+ -0.4; the reference range of BR_kurtosis is 0.7[0.2,1.3];

the MV_mean reference range is 39487.1 +/-13785.4;

the reference range for MV_cv is 32.3[27.5,36.8];

the reference range of MV_Shewness is 0.8[0.6,1.0];

the reference range of MV_kurtosis is 1.3[0.5,2.5];

the reference range of AB_mean is 0.3+ -0.2;

the reference range of ab_cv is 32.7±12.9;

the reference range of AB_skew is 0.7[0.1,1.0];

the reference range of AB_kurtosis is 0.4 < -0.3,1.3 >;

the reference range of IE_mean is 0.7[0.6,0.7]; the reference range of IE_cv is 21.4[19.1,24.6]; the reference range of IE_skew is 1.3[0.8,1.9];

the reference range of IE_kurtosis is 5.3[2.6,10.9];

the reference range of LF is 44023.2[36690.9,59709.5]; the reference range for HF is 1709.9 [14343.5,20944.7]; the reference range of LF_HF is 2.7+/-0.4;

nLF has a reference range of 72.7[70.5,74.7];

the reference range of nHF is 27.3[25.3,29.5]; the reference range for TP is 68646.3[56773.3,90296.6]; the reference range of vLF is 6491.1[5664.9,9438.0]; the reference range for SD1 is 480.3[423.3,558.1];

the reference range for SD2 is 749.8[649.9,871.3];

the reference range of SD2/SD1 is 1.6+/-0.2;

the reference range for SampEn is 1.5[1.4,1.6]; the reference range of area_1_5 is 7.0[6.6,7.5];

the reference range of area_6_20 is 18.2[16.6,19.2]; in the state of the lying position,

the reference range of BR_mean is 15.9[14.5,17.4];

the reference range for BR_cv is 15.6[13.7,18.4];

the reference range of BR_skew is 1.0+ -0.6; the reference range of BR_kurtosis is 2.4[1.2,3.8];

the MV_mean reference range is 19417.5 +/-6476.9;

the reference range for MV_cv is 44.1[35.8,57.3];

the reference range of MV_Shewness is 1.8[1.2,2.5];

the reference range of MV_kurtosis is 6.8[2.6,12.5];

the reference range of AB_mean is 0.6[0.5,0.7];

the reference range of AB_cv is 17.9[15.2,22.3];

the reference range of AB_skew is-0.2+/-0.8; the reference range of AB_kurtosis is 0.3 < -0.5,1.3 >;

the reference range of IE_mean is 0.8[0.7,0.8]; the reference range of IE_cv is 26.6[23.0,31.6]; the reference range of IE_skew is 1.4[1.0,2.1];

the reference range of IE_kurtosis is 4.2[1.7,9.2];

the reference range of LF is 43629.4[31883.0, 6845.2 ]; the reference range for HF is 12454.3[10036.0,18834.6]; the reference range of LF_HF is 3.5[3.2,3.9];

nLF has a reference range of 77.5[76.1,79.6];

the reference range of nHF is 22.5[20.4,23.9]; TP has a reference range of 62917.5[46543.6,97620.1]; the reference range of vLF is 6686.1[4849.2,10321.3]; the reference range for SD1 is 557.7[447.5,739.3];

SD2 has a reference range of 986.3[766.2,1108.4];

the reference range of SD2/SD1 is 1.6+/-0.3;

the reference range for SampEn is 1.4[1.2,1.6]; the reference range of area_1_5 is 6.2[5.1,6.8];

the reference range of area_6_20 is 12.4[10.4,14.0].

Preferably, the step of SQI quality assessment is:

the first step, dividing the respiratory signal into segments of 30s, and then performing the second step;

step two, filtering by using a band-pass filter with the cutoff frequency of 0.1-1Hz, and detecting wave crest and wave trough; then, performing a third step;

thirdly, judging whether the signal meets the normal physiological range of the human body or not, and judging whether the variation coefficient of the respiratory cycle is smaller than a first threshold value or not; judging whether the proportion of the median of the respiratory cycle with the respiratory cycle being more than 1.5 times or less than 0.5 times is less than 15 percent; judging whether the proportion of effective respiration in the segment is larger than a second threshold value; if all the signals are yes, the fourth step is carried out, and if any one of the signals is no, the signal is judged to be a low-quality signal;

step four, evaluating the morphological similarity of the respiration signals, and calculating the correlation coefficient of the respiration signals and the template by respiration through the respiration signal template; then, performing a fifth step;

fifthly, judging whether the average correlation coefficient is larger than a third threshold value; if yes, then high quality signal is obtained; if not, a low quality signal is provided.

The quantitative analysis method and the quantitative analysis device for the 24-hour breathing mode provide objective indexes for quantitative breathing, and provide effective coordinates and scales for 24-hour breathing mode research, so that auxiliary indexes for clinical diagnosis are possible through the quantitative breathing mode.

Drawings

FIG. 1 is a wearable respiratory signal monitoring device SeneEcho as used in the present application;

FIG. 2 is a flow chart of the SQI algorithm for respiratory signal quality assessment in the method and apparatus for quantitative analysis of 24-hour respiratory patterns of the present application;

FIG. 3 is a Bland-Altman plot comparing Ve and Vm breath by breath;

fig. 4 is an example of a signal quality classification result.

Detailed Description

The present application will be described in detail with reference to the accompanying drawings.

A. Wearable device

The respiratory signal monitoring device used in the present application is a medical-grade wearable device SeneEcho, which can provide a single-lead ECG signal, chest and abdomen respiratory signals, and tri-axial acceleration signals, and can be connected with a third party monitoring device such as blood pressure, blood oxygen, etc. in a bluetooth communication manner (fig. 1). The respiratory motion signal sensor adopts RIP technology and consists of two elastic belts of chest and abdomen, and the sampling rate is 25Hz. The body position may be determined by a tri-axial acceleration sensor. SeneEcho allows long-range monitoring for 24 hours and more. SeneEcho has been authenticated by the national drug administration (NMPA) and has been applied to data collection and mining in clinical settings.

B. Tidal volume calibration of wearable devices

Tidal volume, which refers to the amount of air that is normally breathed, is an essential parameter for the breathing pattern. Assuming that the lung volume is approximately equal to the sum of the chest and abdomen volumes, the output of SeneEcho is compared to a portable spirometer as a gold standard to calibrate the tidal volume.

C. Improved respiratory signal quality assessment algorithm

In daily life, when 24-hour breathing pattern monitoring is performed, low quality signal segments, such as noise and artifacts, are inevitably generated due to motion, talking and the wearable device itself. Respiratory signal quality assessment is one of the bases of 24-hour respiratory pattern quantitative analysis and mobile health.

In "An impedance pneumography signal quality index:design, assessment and application to respiratory rate monitoring," (p.h. charlton et al, biomedical signal processing and control, 2021), a Signal Quality Index (SQI) for impedance pneumography was developed to improve respiratory rate monitoring based on reliability and similarity of respiratory signals. The performance was assessed using RRest-vent (P.H. Charlton, "Continuous respiratory rate monitoring to detect clinical deteriorations using wearable sensors," Ph.D. distervation, king's College London, 2017) and the MIMIMIC dataset (A.E. Johnson et al., "Mimic-iii, a freely accessible critical care database," Scientific data, vol.3, no.1, pp.1-9,2016). The respiratory signals acquired by the different devices are different, so in this application we refer to this method but optimize the threshold. We use a band pass filter instead of a low pass filter and generalize the original SQI standard by replacing predefined numbers in the a, b and c positions with adjustable parameters. The SQI flow chart of the present application is shown in FIG. 2.

Breath-by-breath calculation of breath duration, effective breath is defined as the length of the first peak to the last peak in the segment. In this segment, the respiration template is calculated as an average of the respiration waveforms in the segment. In fig. 2, variables a, b, and c are thresholds to be adjusted. To account for low quality signals including those caused by natural body movements and by disease, we constructed a test set using data collected by SeneEcho for 16 patients and 8 healthy persons. The criteria for artificially labeling respiratory signals are listed in table I. The test set was labeled by three independent clinical professionals. Finally, the test set includes 1625 30 second respiratory signal segments, 1323 of which have good signal quality.

Table I signal quality criteria

D. Respiratory signal preprocessing and respiratory modes

The respiratory signal is bandpass filtered using a first order IIR Butterworth filter to remove frequency components above 1Hz and below 0.1 Hz. The breathing pattern is calculated using high quality segments, including time domain, frequency domain, and non-linearities.

Time domain. The importance of respiratory rate and minute ventilation need not be described in detail. The inspiratory duty cycle (inspiratory time/total time) is typically 1/3 of the respiratory cycle, but the patient may vary, especially in patients with upper airway obstruction (m.sowho et al., "Snoring: a source of noise pollution and Sleep apnea predictor," Sleep, 2020). The contribution of the chest and abdomen to tidal volume is a useful indicator for assessing difficult to offline patients (r.priori et al., "Contributions of rib cage (rc) and abdomens (ab) to tidal volume are useful indicators for the assessment of difficult-to-web components," 2012). Thus, the time domain breathing patterns defined in this study include respiratory rate, minute ventilation, ratio of inspiration to expiration time, and abdominal respiratory contribution. The breath-by-breath time domain breathing pattern is defined as follows:

respiratory rate (BR, breath/min): reciprocal of the breathing cycle.

Minute ventilation (MV, au): the product of respiratory rate and tidal volume.

Ratio of inspiration to expiration time (IE): inhalation time/exhalation time.

Abdominal respiratory contribution (AB) Abdominal respiratory amplitude/tidal volume.

We first calculate the breath-by-breath parameters defined above, then interpolate to a sequence of 1 second intervals, and running average through a window width of 10 seconds, an averaging filter that overlaps for 9 seconds, and finally calculate the average, coefficient of Variation (CV), skewness and kurtosis for each parameter.

Frequency domain. The frequency domain method mainly refers to the spectral power densities associated with different frequency bands of the breath-by-breath breathing cycle. High Frequency (HF) 0.15-0.4Hz, low Frequency (LF) 0.04-0.15Hz, ultra low frequency (vLF) <0.04Hz [24]. The total power is the sum of different frequency bands. LF/HF is the ratio of low frequency power to high frequency power. The low frequency power is divided by the total power to obtain a normalized low frequency power (nLF), and the high frequency power is divided by the total power to obtain a normalized high frequency power (nff).

Nonlinear analysis. We calculated poincare plot, sample entropy and multi-scale entropy (MSE) of breath intervals. Poincare is a commonly used non-linear analysis method. The graph is quantized by horizontal and vertical standard deviations SD1 and SD2, respectively. MSE is an improvement to sample entropy (samplen) and is commonly used to evaluate time series complexity. MSE of heart rate sequences has been shown to characterize health. Slope of scale 1-5 (slope_1_5), area under the curve of scale 1-5 (area_1_5), and Area under the curve of scale 6-20 (area_6_20) are common metrics.

E. Breathing patterns in healthy subjects and patients undergoing heart valve surgery

Healthy Subjects (HS). The study was conducted in the China' S civil release army general Hospital, and was approved by the ethical Committee review (ethical number S2018-095-01). Subjects with any cardiopulmonary disease are excluded. All participants in the study signed informed consent. The SensEcho device is worn by the subject at 8 am, and the subject is removed at 8 am the next day without restricting the activity of the patient throughout the course. 70 healthy subjects (36 men and 34 women) who participated in the study were between 24-43 years old, with an average age of 26.4 years and an average BMI of 21.9.

Heart valve disease (VHD) patients. The study was conducted at the university of Sichuan Huaxi hospital and was subjected to ethical committee examination (ethical number: 20211023). Heart valve disease is an important component of cardiovascular disease. To reduce the impact of surgeon surgery on outcome, the study recruited only patients who underwent heart valve surgery on the same medical team. Patients receiving emergency surgery, having a history of cerebrovascular accident, history of cardiac or pulmonary surgery, neuromuscular disease, unstable cardiovascular symptoms and limited physical function are excluded. A total of 76 patients (43 men and 33 women) were eventually enrolled. They were between 32 and 80 years old, with an average age of 67.6 years and an average BMI of 24.1. All participants signed informed consent. Prior to starting the trial, the patient is informed of the content, purpose, method and possible risk of the trial and signs informed consent in voluntary situations. On the day of admission, the patient was fitted with a SensEcho device, and the patient's activities were not restricted throughout the course and removed 10:00 am the next day. All tests were performed and recorded by the same physical therapist and PPCs were assessed daily until 28 days post-surgery.

Breathing patterns in four states. According to clinician recommendations, we divide respiratory monitoring into four states, day (7:00-23:00) and night (the next day 23:00-7:00), and upright and recumbent positions (l.p.d.s. Mendes et al., "Influence of posture, sex, and age on breathing pattern and chest wall motion in healthy subjects," Brazilian journal of physical therapy, vol.24, no.3, pp.240-248,2020), (j.p. mortola, "Breathing around the clock: an overview of the circadian pattern of respiration," European journal of applied physiology, 2004). The motion state was not analyzed for two reasons: for healthy subjects, different types, intensities, and daily exercise durations result in poor comparability between exercise states. For heart valve patients, older age and poor physical condition result in very low activity levels, thus eliminating the need to analyze the movement state separately.

F. Research method

Breathing patterns of healthy subjects. In clinical work, one important reason limiting the development of wearable devices is the lack of a reference range of normal values. The clinician is not aware of what the monitored physiological parameters represent (normal or abnormal). Therefore, it is important to calculate the normal reference range of the breathing pattern and get the clinician's acceptance. For this reason, we calculate the breathing pattern of healthy subjects as the normal reference range of breathing patterns, so the clinician has confidence in exploring anomalies in the patient's breathing pattern. It is natural that there is a significant difference in breathing patterns between healthy subjects and patients. Because of the significant age differences in these two cohorts, we did not make such comparisons between healthy subjects and heart valve patients of our interest, and the comparison was unfair under different health conditions of the subjects.

Breathing patterns of heart valve patients. The occurrence of PPCs results in significant increases in post-operative mortality and hospitalization costs, and thus, the increased residence time and hospitalization time of the Intensive Care Unit (ICU), and thus, the pre-operative risk assessment associated with PPCs of patients receiving heart valve surgery has an important role in planning clinical treatments, determining prognosis, predicting complications, assessing treatment outcome and assessing medical resource requirements (j. Iba-nez et al, "Long-term mortality after pneumonia in cardiac surgery patients: adaptive-matched analysis," Journal of intensive care medicine, vol.31, no.1, pp.34-40,2016), (m. Garc '1a-Delgado, i. Navarete-S' ancez, and m. Colmeno, "Preventing and managing perioperative pulmonary complications following cardiac surgery," Current Opinion in Anesthesiology, 2014). The occurrence of PPCs has been shown to correlate with pre-operative cardiopulmonary function test results. Thus, we performed a sub-study comparing differences and similarities between the non-PPCs group and the PPCs group to find a correlation between 24 hour breathing patterns and postoperative complications in heart valve operated patients. If a 24 hour preoperative monitoring of the respiratory pattern can determine a high risk patient for PPC, the incidence of PPCs can be reduced by preoperative home pre-rehabilitation.

Statistical methods. We analyzed the breathing patterns of healthy and VHD groups using statistical methods. For the variables in section II-D, the normal continuous variable is expressed as the average (standard deviation) and the non-normal continuous variable is expressed as the median [ quartile range]As shown in tables IV and VI. The comparison between normal data sets uses a double-sample t-test, and the comparison between bias data sets uses a non-parametric rank sum test. Classification variable is expressed as n (%), and χ is used for comparison between groups ² The differences between groups are considered statistically significant by testing and Fisher exact probability method, p < 0.05.

Results

Tidal volume calibration

Table II gives the absolute errors of Ve and Vm calculated breath by breath and the correlation coefficients. The average absolute error was (-0.29.+ -. 83.20) ml and the average correlation coefficient was 0.958. Fig. 3 shows a Bland-Altman plot comparing Ve and Vm breath by breath. The x-axis represents the average tidal volume obtained for both methods and the y-axis represents the error value between Ve and Vm. Wherein 97.0% of the dots are distributed within a mean + -1.96 SD (-0.29 + -1.08 mL) range. Statistical analysis was performed on all breath-by-breath tidal volumes (2339 times) obtained by both methods, with a relative tidal volume error of less than 20% of 96.71% (2262 times).

TABLE II absolute error and correlation coefficient of tidal volume per tidal volume

SQI Performance

The performance of SQI on the test set using different thresholds is shown in Table III. The results show that different threshold combinations have a greater impact on model performance. For example, the accuracy rate can be changed in a range of 79.88% -92.37%. "a=0.35, b=0.5, c=0.7" is the same as "a=0.35, b=0.5, c=0.75" in terms of accuracy and highest in both combinations, the latter being more specific. In view of the higher specificity, signal segments with poor signal quality can be better identified, so that the accuracy of the breathing pattern calculation is improved, and 'a=0.35, b=0.5 and c=0.75' are taken as optimal thresholds and used in subsequent analysis. Fig. 4 shows an example of a signal quality classification result.

Table III performance of SQI on test set. The itutic font represents the original setting. The best performance is indicated in bold.

Breathing patterns in healthy subjects

One important reason for restricting the development of wearable devices is that the clinician does not know the explicit meaning and reference range of the wearable device monitoring index. In our study, all breathing pattern parameters extracted are interpretable and give a reference range for healthy subjects, which may facilitate clinical use. The breathing patterns of healthy subjects in 4 states (day, night, upright and recumbent) were calculated and the results are shown in table IV.

Table IV breathing patterns of healthy people

In the table IV of the present description,

BR represents the respiratory rate (BR, times/min), which is the inverse of the respiratory cycle; br_mean is the respiratory rate average; BR_cv is the respiratory rate variation coefficient; BR_skew is the respiratory rate bias; br_kurtosis is respiratory rate kurtosis.

MV represents minute ventilation, which is the product of respiratory rate and tidal volume; MV_mean is the minute ventilation average; MV_cv is a minute ventilation variable coefficient; MV_skewness is minute ventilation bias; MV_kurtosis is minute ventilation kurtosis.

AB represents the abdominal respiration contribution, which is the abdominal respiration amplitude/tidal volume; AB mean is the average of the abdominal respiration contribution; ab_cv ventral respiratory contribution coefficient of variation; ab_skew is the ventral respiratory contribution degree bias; ab_kurtosis is the kurtosis of the abdominal respiration contribution.

IE represents the ratio of inspiration to expiration time, which is inspiration time/expiration time; ie_mean is the average value of the ratio of inspiration to expiration times; ie_cv is the coefficient of variation of the ratio of inspiration to expiration times; IE_skew is the bias of the ratio of inspiration to expiration times; ie_kurtosis is the kurtosis of the ratio of inspiration to expiration times.

LF represents low frequency power, which is 0.04-0.15 Hz.

HF means high frequency power, 0.15-0.4 Hz.

Lf_hf (or LF, HF) represents the ratio of low frequency to high frequency, which is the ratio of low frequency power to high frequency power.

nLF the normalized low frequency power is the low frequency power divided by the total power.

nff represents normalized high frequency power, which is the high frequency power divided by the total power.

TP represents the total power, which is the sum of the different band powers.

vLF denotes ultra-low frequency power, being <0.04 Hz.

SD1 represents the horizontal standard deviation, which is

Wherein, RR interval is respiratory interval, SD () is standard deviation calculation formula.

SD2 represents the vertical standard deviation, which is

SD2/SD1 is the ratio of SD2 to SD 1.

Samplen denotes a sample entropy, which is samplen=ln (a (m, r)) -ln (a (m+1, r)),

wherein A (m, r) represents the number of similar patterns when the template length is m, and r is a similarity criterion.

Slope_1_5 is the Slope of the multiscale entropy of the heart rate sequence at scales 1-5.

Area_1_5 is the Area under the curve of scale 1-5 of the multiscale entropy of the heart rate sequence.

Area_6_20 is the Area under the curve of the scale 6-20 of the multiscale entropy of the heart rate sequence.

Breathing pattern of heart valve surgical patient

(1) Clinical features of both non-PPCs and PPCs patients were analyzed. PPCs are known to be a significant cause of poor post-operative prognosis, mortality and increased medical costs in patients with heart valve surgery. Therefore, risk assessment of preoperative PPC has become an important aspect of patient management. PPC occurred in 15 of 76 patients. The clinical characteristics of both groups of patients are shown in Table V. The diabetic prevalence was higher in PPC group and the surgical time was longer than in non-PPC.

(2) Breathing patterns of heart valve patients. The PPC group had a smaller coefficient of variation in the respiratory rate in the prone position than the non-PPCs group (see table VI). The respiratory rate kurtosis and the abdominal respiratory contribution ratio of the PPCs group were lower during the day. This indicates that the respiratory rate distribution and abdominal contribution of the PPC group is more diffuse during the day. At night, the variability of the breathing rate of the PPCs group, as well as SD2, was lower, similar to the results in the prone position.

Clinical characterization of two groups of patients in Table V

Table VI typical breathing patterns for heart valve patients

Wearable devices have been widely developed in the last decades. However, one challenge is that when monitoring the normal state of a subject, signal quality is poor, assessment is inaccurate, and the rate of false positives is high due to difficult-to-control environments and behaviors (l.lu et al., "Wearable health devices in health care: narrative systematic review," JMIR mHealth and uHealth, vol.8, no.11,2020). To alleviate this problem, we use a signal quality assessment method that relies on peak/trough detection. This lightweight feature is more suitable for real-time deployment in a clinical setting. In addition, we have improved the SQI standard, making it more suitable for the Chinese population.

This study identified potential predictors by quantifying the 24 hour pre-operative respiratory pattern of heart valve operated patients, and identified high risk populations of PPCs and those requiring pre-operative rehabilitation.

Unless defined otherwise, all technical and/or scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The materials, methods, and examples mentioned in this application are illustrative only and not intended to be limiting.

Although the present invention has been described in connection with specific embodiments thereof, those skilled in the art will appreciate that various substitutions, modifications and changes may be made without departing from the spirit of the invention.

Claims

1. A quantitative analysis method of 24-hour breathing mode includes the steps that breathing signals of a subject are obtained through wearable equipment, SQI quality evaluation is conducted on the breathing signals, and high-quality breathing signals are obtained; the following parameters are obtained from the high quality respiratory signal:

BR_mean、BR_cv、BR_skewness、BR_kurtosis、

MV_mean、MV_cv、MV_skewness、MV_kurtosis、

AB_mean、AB_cv、AB_skewness、AB_kurtosis、

IE_mean、IE_cv、IE_skewness、IE_kurtosis、

LF、HF、LF_HF、nLF、nHF、TP、vLF、

SD1、SD2、SD2/SD1、

SampEn、Area_1_5、Area_6_20；

2. The method for quantitative analysis of 24-hour breathing patterns according to claim 1, wherein:

dividing the respiration of the subject into four states, namely a daytime state, a nighttime state, a static upright state and a lying state;

the reference ranges of the parameters are different in different states.

3. The method for quantitative analysis of 24-hour breathing patterns according to claim 2, wherein:

in the daytime state of the vehicle,

the reference range of BR_mean is 21.3+ -3.0;

the reference range of BR_cv is 22.1+ -3.5;

the reference range of BR_skew is 0.8+ -0.4;

the reference range of BR_kurtosis is 0.4[0.2,1.2];

the MV_mean reference range is 44700.8 +/-16377.4;

the reference range for MV_cv is 49.1[43.2,58.5];

the reference range of MV_Shewness is 1.3+ -0.6;

the reference range of MV_kurtosis is 2.1[0.7,4.2];

the reference range of AB_mean is 0.4+/-0.2;

the reference range for AB_cv is 39.5[32.9,46.6];

the reference range of AB_skew is 0.6[0.1,1.1];

the reference range of AB_kurtosis is-0.2 < -0.7,1.2 >;

the reference range of IE_mean is 0.7+ -0.1;

the reference range of IE_cv is 25.1[22.6,28.6];

the reference range of IE_skew is 1.2[0.9,1.7];

the reference range of IE_kurtosis is 3.2[2.1,6.1];

nLF with a reference range of 70.4 + -3.3;

the reference range of nHF is 29.6+/-3.3;

SD2 has a reference range of 969.9[865.9,1107.8];

the reference range of SD2/SD1 is 1.8[1.7,2.0];

the reference range for SampEn is 1.4[1.3,1.6];

the reference range of area_1_5 is 6.6[6.1,7.1];

the reference range of area_6_20 is 16.7[15.3,17.5];

in the night-time state,

the reference range of BR_mean is 16.3[15.1,17.8];

the reference range of BR_cv is 19.8+ -5.2;

the reference range of BR_skew is 1.5+ -0.6;

the reference range of BR_kurtosis is 3.3[1.9,6.6];

the MV_mean reference range is 22514.3 +/-8087.3; the reference range for MV_cv is 64.2[53.3,76.5];

the reference range of MV_Shewness is 2.4[1.8,3.2];

the reference range of MV_kurtosis is 9.5[4.3,15.0];

the reference range of AB_mean is 0.6[0.5,0.7];

the reference range for AB_cv is 23.2[18.2,29.0];

the reference range of AB_skew is-0.5 [ -1.1, -0.0]; the reference range of AB_kurtosis is 0.5 < -0.3,1.8 >;

the reference range of IE_mean is 0.8[0.7,0.8];

the reference range of IE_cv is 28.1[24.2,32.2];

the reference range of IE_skew is 1.4[1.0,2.0];

the reference range of IE_kurtosis is 4.6[1.7,8.2];

nLF has a reference range of 76.9[74.1,78.8];

the reference range of nHF is 23.1[21.2,25.9];

the reference range for SD2 is 1067.6[885.3,1344.3];

the reference range of SD2/SD1 is 1.8+/-0.4;

the reference range for SampEn is 1.4[1.2,1.5];

the reference range of area_1_5 is 5.7[4.9,6.5];

the reference range of area_6_20 is 11.4[9.4,12.8];

in the state of resting and standing upright,

the reference range of BR_mean is 19.8+ -2.7; the reference range for BR_cv is 14.8[13.2,16.5];

the reference range of BR_skew is 0.3+ -0.4;

the reference range of BR_kurtosis is 0.7[0.2,1.3];

the MV_mean reference range is 39487.1 +/-13785.4; the reference range for MV_cv is 32.3[27.5,36.8];

the reference range of MV_Shewness is 0.8[0.6,1.0]; the reference range of MV_kurtosis is 1.3[0.5,2.5];

the reference range of AB_mean is 0.3+ -0.2;

the reference range of ab_cv is 32.7±12.9;

the reference range of AB_skew is 0.7[0.1,1.0];

the reference range of AB_kurtosis is 0.4 < -0.3,1.3 >;

the reference range of IE_mean is 0.7[0.6,0.7];

the reference range of IE_cv is 21.4[19.1,24.6];

the reference range of IE_skew is 1.3[0.8,1.9];

the reference range of IE_kurtosis is 5.3[2.6,10.9];

nLF has a reference range of 72.7[70.5,74.7];

the reference range of nHF is 27.3[25.3,29.5];

the reference range for TP is 68646.3[56773.3,90296.6]; the reference range of vLF is 6491.1[5664.9,9438.0]; the reference range for SD1 is 480.3[423.3,558.1];

the reference range for SD2 is 749.8[649.9,871.3];

the reference range of SD2/SD1 is 1.6+/-0.2;

the reference range for SampEn is 1.5[1.4,1.6];

the reference range of area_1_5 is 7.0[6.6,7.5]; the reference range of area_6_20 is 18.2[16.6,19.2]; in the state of the lying position,

the reference range of BR_mean is 15.9[14.5,17.4];

the reference range for BR_cv is 15.6[13.7,18.4];

the reference range of BR_skew is 1.0+ -0.6;

the reference range of BR_kurtosis is 2.4[1.2,3.8];

the MV_mean reference range is 19417.5 +/-6476.9; the reference range for MV_cv is 44.1[35.8,57.3];

the reference range of MV_Shewness is 1.8[1.2,2.5]; the reference range of MV_kurtosis is 6.8[2.6,12.5]; the reference range of AB_mean is 0.6[0.5,0.7];

the reference range of AB_cv is 17.9[15.2,22.3];

the reference range of AB_skew is-0.2+/-0.8;

the reference range of AB_kurtosis is 0.3 < -0.5,1.3 >;

the reference range of IE_mean is 0.8[0.7,0.8];

the reference range of IE_cv is 26.6[23.0,31.6];

the reference range of IE_skew is 1.4[1.0,2.1];

the reference range of IE_kurtosis is 4.2[1.7,9.2];

nLF has a reference range of 77.5[76.1,79.6];

the reference range of nHF is 22.5[20.4,23.9];

TP has a reference range of 62917.5[46543.6,97620.1]; the reference range of vLF is 6686.1[4849.2,10321.3]; the reference range for SD1 is 557.7[447.5,739.3];

SD2 has a reference range of 986.3[766.2,1108.4];

the reference range of SD2/SD1 is 1.6+/-0.3;

the reference range for SampEn is 1.4[1.2,1.6];

the reference range of area_1_5 is 6.2[5.1,6.8];

the reference range of area_6_20 is 12.4[10.4,14.0].

4. The method for quantitative analysis of 24-hour breathing patterns according to claim 1, wherein:

the SQI quality evaluation comprises the following steps:

5. A quantitative analysis device for a 24-hour breathing mode, which comprises a signal acquisition unit, an SQI quality evaluation unit, a parameter acquisition unit and a parameter comparison unit:

BR_mean、BR_cv、BR_skewness、BR_kurtosis、

MV_mean、MV_cv、MV_skewness、MV_kurtosis、

AB_mean、AB_cv、AB_skewness、AB_kurtosis、

IE_mean、IE_cv、IE_skewness、IE_kurtosis、

LF、HF、LF_HF、nLF、nHF、TP、vLF、

SD1、SD2、SD2/SD1、

SampEn、Area_1_5、Area_6_20；

6. The quantitative analysis device for 24-hour breath pattern according to claim 5, wherein:

the parameter acquisition unit divides the respiration of the subject into four states, namely a daytime state, a nighttime state, a static upright state and a lying state;

the reference ranges of the parameters employed by the parameter comparison unit are different in different states.

7. The quantitative analysis device for 24-hour breath pattern according to claim 6, wherein:

in the daytime state of the vehicle,

the reference range of BR_mean is 21.3+ -3.0;

the reference range of BR_cv is 22.1+ -3.5;

the reference range of BR_skew is 0.8+ -0.4;

the reference range of BR_kurtosis is 0.4[0.2,1.2];

the MV_mean reference range is 44700.8 +/-16377.4;

the reference range for MV_cv is 49.1[43.2,58.5];

the reference range of MV_Shewness is 1.3+ -0.6;

the reference range of MV_kurtosis is 2.1[0.7,4.2];

the reference range of AB_mean is 0.4+/-0.2;

the reference range for AB_cv is 39.5[32.9,46.6];

the reference range of AB_skew is 0.6[0.1,1.1];

the reference range of AB_kurtosis is-0.2 < -0.7,1.2 >;

the reference range of IE_mean is 0.7+ -0.1;

the reference range of IE_cv is 25.1[22.6,28.6];

the reference range of IE_skew is 1.2[0.9,1.7];

the reference range of IE_kurtosis is 3.2[2.1,6.1];

the reference range of LF is 56887.3[48891.5,67446.2];

the reference range for HF is 25005.8[20613.1,27915.5]; the reference range of LF_HF is 2.4[2.2,2.7];

nLF with a reference range of 70.4 + -3.3;

the reference range of nHF is 29.6+/-3.3;

SD2 has a reference range of 969.9[865.9,1107.8];

the reference range of SD2/SD1 is 1.8[1.7,2.0];

the reference range for SampEn is 1.4[1.3,1.6];

the reference range of area_1_5 is 6.6[6.1,7.1];

the reference range of area_6_20 is 16.7[15.3,17.5];

in the night-time state,

the reference range of BR_mean is 16.3[15.1,17.8];

the reference range of BR_cv is 19.8+ -5.2;

the reference range of BR_skew is 1.5+ -0.6;

the reference range of BR_kurtosis is 3.3[1.9,6.6];

the reference range of MV_Shewness is 2.4[1.8,3.2];

the reference range of MV_kurtosis is 9.5[4.3,15.0];

the reference range of AB_mean is 0.6[0.5,0.7];

the reference range for AB_cv is 23.2[18.2,29.0];

the reference range of IE_mean is 0.8[0.7,0.8]; the reference range of IE_cv is 28.1[24.2,32.2];

the reference range of IE_skew is 1.4[1.0,2.0];

the reference range of IE_kurtosis is 4.6[1.7,8.2];

nLF has a reference range of 76.9[74.1,78.8];

the reference range of nHF is 23.1[21.2,25.9];

the reference range for SD2 is 1067.6[885.3,1344.3];

the reference range of SD2/SD1 is 1.8+/-0.4;

the reference range for SampEn is 1.4[1.2,1.5];

the reference range of area_1_5 is 5.7[4.9,6.5];

the reference range of area_6_20 is 11.4[9.4,12.8];

in the state of resting and standing upright,

the reference range of BR_mean is 19.8+ -2.7;

the reference range for BR_cv is 14.8[13.2,16.5];

the reference range of BR_skew is 0.3+ -0.4;

the reference range of BR_kurtosis is 0.7[0.2,1.3];

the reference range of MV_Shewness is 0.8[0.6,1.0];

the reference range of MV_kurtosis is 1.3[0.5,2.5];

the reference range of AB_mean is 0.3+ -0.2; the reference range of ab_cv is 32.7±12.9;

the reference range of AB_skew is 0.7[0.1,1.0];

the reference range of AB_kurtosis is 0.4 < -0.3,1.3 >;

the reference range of IE_mean is 0.7[0.6,0.7];

the reference range of IE_cv is 21.4[19.1,24.6];

the reference range of IE_skew is 1.3[0.8,1.9];

the reference range of IE_kurtosis is 5.3[2.6,10.9];

nLF has a reference range of 72.7[70.5,74.7];

the reference range of nHF is 27.3[25.3,29.5];

the reference range for SD2 is 749.8[649.9,871.3];

the reference range of SD2/SD1 is 1.6+/-0.2;

the reference range for SampEn is 1.5[1.4,1.6];

the reference range of area_1_5 is 7.0[6.6,7.5];

the reference range of BR_mean is 15.9[14.5,17.4];

the reference range for BR_cv is 15.6[13.7,18.4];

the reference range of BR_skew is 1.0+ -0.6;

the reference range of BR_kurtosis is 2.4[1.2,3.8];

the reference range of MV_Shewness is 1.8[1.2,2.5];

the reference range of MV_kurtosis is 6.8[2.6,12.5];

the reference range of AB_mean is 0.6[0.5,0.7];

the reference range of AB_cv is 17.9[15.2,22.3];

the reference range of AB_skew is-0.2+/-0.8;

the reference range of AB_kurtosis is 0.3 < -0.5,1.3 >;

the reference range of IE_mean is 0.8[0.7,0.8];

the reference range of IE_cv is 26.6[23.0,31.6];

the reference range of IE_skew is 1.4[1.0,2.1];

the reference range of IE_kurtosis is 4.2[1.7,9.2];

the reference range of LF is 43629.4[31883.0, 6845.2 ];

the reference range for HF is 12454.3[10036.0,18834.6];

the reference range of LF_HF is 3.5[3.2,3.9];

nLF has a reference range of 77.5[76.1,79.6];

the reference range of nHF is 22.5[20.4,23.9];

TP has a reference range of 62917.5[46543.6,97620.1];

the reference range of vLF is 6686.1[4849.2,10321.3];

the reference range for SD1 is 557.7[447.5,739.3];

SD2 has a reference range of 986.3[766.2,1108.4];

the reference range of SD2/SD1 is 1.6+/-0.3;

the reference range for SampEn is 1.4[1.2,1.6];

the reference range of area_1_5 is 6.2[5.1,6.8];

the reference range of area_6_20 is 12.4[10.4,14.0].

8. The quantitative analysis device for 24-hour breath pattern according to claim 5, wherein: the SQI quality evaluation unit performs a step of performing an SQI quality evaluation;