WO2021135672A1 - 一种用于判断睡眠呼吸暂停的信号检测方法及系统 - Google Patents

一种用于判断睡眠呼吸暂停的信号检测方法及系统 Download PDF

Info

Publication number
WO2021135672A1
WO2021135672A1 PCT/CN2020/128426 CN2020128426W WO2021135672A1 WO 2021135672 A1 WO2021135672 A1 WO 2021135672A1 CN 2020128426 W CN2020128426 W CN 2020128426W WO 2021135672 A1 WO2021135672 A1 WO 2021135672A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
sleep
signals
bcg
sample
Prior art date
Application number
PCT/CN2020/128426
Other languages
English (en)
French (fr)
Inventor
张涵
朱玮玮
叶颂斌
余宝贤
Original Assignee
华南师范大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华南师范大学 filed Critical 华南师范大学
Priority to US17/790,445 priority Critical patent/US20230043406A1/en
Publication of WO2021135672A1 publication Critical patent/WO2021135672A1/zh
Priority to US18/480,906 priority patent/US20240023886A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
    • A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
    • A61B5/1102—Ballistocardiography
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/48—Other medical applications
    • A61B5/4806—Sleep evaluation
    • A61B5/4818—Sleep apnoea
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/20—Ensemble learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients

Definitions

  • the invention relates to the field of sleep breathing signal research, in particular to a signal detection method and system for judging sleep apnea.
  • the embodiments of the present invention provide a signal detection method and system for judging sleep apnea.
  • a signal detection method for determining sleep apnea which includes the following steps:
  • the effective sign signal set is input into the sleep breathing detection model for signal processing to obtain the probability data of the user's apnea during sleep.
  • the signal detection method for judging sleep apnea described in this embodiment uses multi-dimensional morphological features to perform feature training on the initial model of the classifier to obtain a sleep respiration detection model, which makes the performance of the sleep respiration detection model more robust and can obtain more Accurate data on the probability of the user's apnea occurring during sleep, so that the user or doctor can accurately determine whether the user's apnea event occurs during the sleep.
  • the method of structurally processing the vital sign signals of the user during sleep to remove invalid signals, and obtaining a valid sign signal set includes the following steps:
  • the vital sign signals after the invalid signal interval are removed are reasonably spliced to obtain the effective sign signal set after the interference is removed.
  • the method of extracting and using the multi-dimensional morphological features of the sleep breathing sample signal to perform feature training on the initial model of the classifier to obtain the sleep breathing detection model includes the following steps:
  • Extracting a set of multi-dimensional morphological features of the BCG sample signal in a fixed time scale the multi-dimensional morphological features including: low-frequency features, peak features, area features, power spectrum features, and nonlinear features;
  • the method for extracting the BCG sample signal in the effective sleep breathing sample signal is:
  • screening out BCG sample signals in a fixed time scale includes the following steps:
  • the method for inputting the extracted multi-dimensional morphological feature set of the BCG sample signal into the integrated learning model for feature optimization includes the following steps:
  • Random up and down permutation of specific columns in the multi-dimensional morphological feature set and input the multi-dimensional morphological feature set after the up and down permutation of the specific column into the tree model for sample feature training to obtain the second error value;
  • a preset empirical threshold is used to delete corresponding morphological features whose absolute value of the difference between the first error value and the second error value is less than the preset empirical threshold to obtain an optimized feature set;
  • the initial model of the classifier includes an LR classifier, an SVM classifier, an RF classifier, and an AdaBoost classifier.
  • a signal detection system for determining sleep apnea including:
  • Vital sign signal acquisition device used to collect the vital sign signals of the user during sleep
  • Memory used to store programs
  • the processor is configured to execute the program stored in the memory to implement the method described in any one of the above items.
  • the signal detection system for judging sleep apnea collects the mysterious physical sign signal of the user during sleep through a portable, non-contact vital sign signal acquisition device, which brings a better test experience to the user. It will affect the user's normal sleep, and the signal detection system can filter the vital sign signals of the user while sleeping, and the signal processing and analysis are more accurate, so that the user or doctor can make accurate judgments about the apnea event of the user while sleeping.
  • FIG. 1 is a schematic flowchart of a signal detection method for determining sleep apnea according to an embodiment of the present invention
  • step 2 is a schematic flowchart of step 2 of the signal detection method for determining sleep apnea according to an embodiment of the present invention
  • FIG. 3 is a schematic flowchart of step 3 of the signal detection method for determining sleep apnea according to an embodiment of the present invention.
  • FIG. 1 is a schematic flowchart of a signal detection method for determining sleep apnea according to an embodiment of the present invention.
  • S2 Structurally process the vital sign signals of the user during sleep to remove invalid signals to obtain an effective sign signal set
  • S4 Input the effective sign signal set into the sleep breathing detection model to perform signal processing to obtain the probability data of the user's apnea during sleep.
  • the signal detection method for judging sleep apnea extracts the multi-dimensional morphological features of the sleep breathing sample signal, and uses the multi-dimensional morphological features to perform feature training on the initial model of the classifier to obtain the sleep respiration detection model.
  • the effective physical sign signal set after chemical processing is input into the sleep breathing detection model, and the signal processing is performed to obtain more accurate data on the probability of apnea of the user during sleep, so that the user or doctor can accurately determine whether the user has an apnea event during sleep.
  • FIG. 2 is a schematic flowchart of step 2 of the signal detection method for determining sleep apnea according to an embodiment of the present invention.
  • step S2 the step of structurally processing the vital sign signals of the user during sleep to remove invalid signals to obtain a valid sign signal set, further includes the following steps:
  • a time scale T can be defined according to the statistical characteristics of Gaussian white noise, the sleep breathing signal in this time scale T can be analyzed, and the mathematical expectation of the sleep breathing signal in the time scale can be obtained.
  • Power spectral density, auto-correlation and signal amplitude when the mathematical expectation is approximately zero, the power spectral density is approximately constant, the time-domain signal auto-correlation is approximately an impact, and the signal amplitude is less than the preset fixed threshold, the judgment at this time
  • the sleep breathing signal is a signal for the user to get out of bed, and the signal is removed.
  • the method includes the general motion signal removal method and the small body motion signal removal method.
  • the general motion signal removal method is: in a fixed time scale T, the general motion start and end time is determined according to the preset limit threshold of the signal, and the amplitude limit threshold is adjusted to determine the general motion signal, and the general motion signal is removed.
  • the small body motion signal removal method is: using Hilbert change to calculate the envelope function mu(t) in the sleep breathing signal, and calculating the ratio of the maximum value max ⁇ to the minimum value min ⁇ in the envelope function, And the ratio of the maximum value max ⁇ to the mean value mean ⁇ , when max ⁇ mu(t) ⁇ /min ⁇ mu(t) ⁇ >p1 and max ⁇ mu(t) ⁇ /mean ⁇ mu(t) ⁇ > At p2, where p1 and p2 are empirical thresholds respectively, it is judged that there is a small body motion signal in this time scale, and the amplitude limit threshold is adjusted to determine the small body motion signal, and the small body motion signal is removed.
  • the signal-to-noise ratio of the output signal is too low, which is not meaningful for analysis.
  • the approximate periodicity originally possessed by the ballistic cardiogram and the breathing signal is submerged by noise, so the invalid signal interval needs to be removed.
  • the method for determining the validity of the signal is: through empirical mode decomposition or wavelet transformation, analyzing the cyclostationary characteristics and autocorrelation of the vital sign signals in the fixed time scale T in different frequency intervals, and determining the time scale Whether the sleep breathing signal is valid, and remove the corresponding interval of the invalid signal.
  • S24 Reasonably splicing the vital sign signals after removing the invalid signal interval to obtain an effective sign signal set after interference removal.
  • the following steps are further included: the first-order statistics and the second-order statistics of the signals in adjacent intervals are respectively counted, and the first-order statistics and the second-order statistics of the signals in the adjacent intervals are calculated separately.
  • the data is directly merged and spliced; when the first-order statistics and second-order statistics of the signals in the adjacent interval are not less than the preset fixed threshold, the two segments of signals are classified and judged separately.
  • FIG. 3 is a schematic flowchart of step 3 of the signal detection method for determining sleep apnea according to an embodiment of the present invention.
  • step S3 the method of extracting and performing feature training on the initial model of the classifier through the multi-dimensional morphological features of the sleep breathing sample signal to obtain the sleep breathing detection model includes the following steps:
  • S31 Perform structured processing on the sleep breathing sample signal to remove invalid signals, and obtain a set of valid sleep breathing sample signals;
  • S33 Extract a set of multi-dimensional morphological features of the BCG sample signal in a fixed time scale, where the multi-dimensional morphological features include: low-frequency features, peak features, area features, power spectrum features, and nonlinear features;
  • S35 Input the steady-state feature set into the initial models of multiple classifiers to perform feature classification training to obtain a sleep breathing detection model.
  • step S32 is:
  • the method for screening BCG sample signals in a fixed time scale is:
  • step S33 the method for extracting multi-dimensional morphological features includes the following methods.
  • the information between the J peak and the K peak is extracted, and the upper and lower envelope functions in the time window are obtained through multiple spline interpolation, and the The upper and lower envelope functions are subjected to empirical mode decomposition to extract the low-frequency components in the upper and lower envelope functions, which are defined as Eu(t) and Ed(t). Specifically, the upper and lower envelopes cover the peaks J and K of the BCG signal.
  • the upper envelope is defined as a function of mu(t)
  • the lower envelope is defined as a function of md(t)
  • mu(t) and md( t) Perform empirical mode decomposition, extract the low-frequency part of the upper and lower envelope functions and define them as Eu(t) and Ed(t).
  • the low-frequency component characteristics can truly reflect the original volatility of the signal. Therefore, the volatility and complexity of Eu(t) and Ed(t) can be used as one of the characteristics of apnea judgment.
  • the variance, standard deviation, and standard deviation of the set ⁇ CJ(i) ⁇ and ⁇ CK(i) ⁇ are also calculated. Kurtosis, slope, etc., and use the calculated variance, standard deviation, kurtosis, and slope as features for identifying apnea;
  • the first-order difference between the adjacent data in the ⁇ CJ(i) ⁇ and ⁇ CK(i) ⁇ sets is calculated, and a new set ⁇ CJ(i) ⁇ and ⁇ CK(i) ⁇ is reconstructed, and Further solve the second-order difference, construct new sets ⁇ 2CJ(i) ⁇ and ⁇ 2CK(i) ⁇ , perform one-dimensional and two-dimensional statistics on the first-order and second-order differences of the data set, and calculate the numerical variance And standard deviation, the numerical variance and standard deviation are used as the characteristics of identifying apnea decision.
  • a fixed-length Fourier transform is calculated for each complete BCG signal, and the ratio of the high and low frequency of the power spectrum density of adjacent BCG signals is quantified, and the low-frequency volatility of the power spectrum of all BCG signals in a fixed time scale is regarded as a feature of the apnea decision .
  • the original signal under a fixed time scale is down-converted, and then the sample entropy is calculated to obtain the sample entropy.
  • Entropy is a measure of the uncertainty of random variables. The greater the uncertainty, the greater the entropy value.
  • the sample entropy is used.
  • the sample entropy is similar to the approximate entropy in the calculation of other entropy values. It has two advantages: first, the sample entropy processing operation does not need to consider the length of the data; secondly, the sample entropy has a lot of Good consistency.
  • sample entropy is well used in evaluating the degree of disorder with time-series physiological signals and in diagnosing pathology, so as to distinguish the difference in value between pause segments and normal respiratory segments.
  • step S34 the method of inputting the extracted multi-dimensional morphological feature set of the BCG sample signal into the integrated learning model for feature optimization includes the following steps:
  • Input the multi-dimensional morphological feature set into the tree model for sample feature training to obtain the first error value, that is, for any tree model Qi, i 1, 2,..., N, where N is the number of trees in the model Number, training is performed based on the feature samples extracted for the first time, and the first error e1 is obtained;
  • the preset empirical threshold is to delete the corresponding morphological features whose absolute value of the difference between the first error value and the second error value is less than the preset empirical threshold to obtain an optimized feature set; that is, to set an empirical threshold ⁇ , when Feature importance
  • step S4 the effective sign signal set is input into the sleep breathing detection model to perform signal processing to obtain the probability data of the apnea of the user during sleep.
  • the classification decision method is: with the aid of the optimized feature set, N (N>1) classifiers are used to classify apneas, the classifiers include LR, SVM, RF, AdaBoost, etc.; combined with the output results of different classifiers The error has a certain degree of statistical independence.
  • the output results of each classifier are weighted and voted, that is, the ratios are combined, and the final output is a binary classification judgment for apnea events, so as to facilitate the discovery of non-frequent short-term abnormal fragments in long-term physical signs.
  • the weighting coefficient is in inverse proportion to the error coefficient of the training set.
  • a quantitative analysis based on different time scales is also performed on the probability data to determine the probability data Whether the segment signal in the corresponding time scale has volatility decline, if the segment signal in the corresponding time scale has volatility decline, it means that the segment signal is a sleep apnea signal, which can verify the probability data obtained by the sleep breathing detection model. accurate.
  • the volatility obtained from a smaller time scale analysis is a small volatility
  • the volatility obtained from a larger time scale analysis is a larger volatility
  • the MAD value can describe the degree of fluctuation of the signal very well.
  • analyzing the volatility of the sleep apnea signal segment at different time scales can reflect whether the apnea signal segment has a significant decrease in volatility relative to the normal breathing segment.
  • the ratio of the minimum volatility to the average volatility is less than 0.2, that is, when c T ⁇ 0.2, it indicates that the apnea signal segment has a significant decline in volatility relative to the normal respiratory segment, which can verify the accuracy of the classification.
  • the multi-scale normalized variance analysis method can also better describe the signal steady state and the degree of alternation.
  • the multi-scale normalized variance analysis method can calculate the multi-scale apnea signal segment and the normal respiratory signal.
  • the ratio of normalized variances between segments can also verify the accuracy of classification.
  • the embodiment of the present invention discloses a signal detection method for judging sleep apnea.
  • the vital sign signals are collected through piezoelectric sensors, and the collected vital signs signals are structured to remove body motion noise and other noises, and then
  • the effective sign signal set is input into the sleep breathing detection model trained based on multi-dimensional morphological features for signal processing to obtain the probability data of the user's apnea during sleep, so that the user or doctor can more accurately determine whether the user has an apnea event during sleep , And the time period during which the apnea event occurred.
  • This method has practical engineering reference significance for future pre-checks such as sleep apnea at home outside the hospital.
  • the embodiment of the present invention also discloses a signal detection system for judging sleep apnea, which includes: a vital sign signal acquisition device for collecting vital sign signals of a person to be tested; a memory for storing programs; a processor for using To implement the above-mentioned method by executing the program stored in the memory.
  • the vital sign signal acquisition device is a piezoelectric sensor module.
  • the piezoelectric sensor module When the signal is collected, the piezoelectric sensor module only needs to be placed under the user's head to observe the user’s sleep and breathing for a long time. The user can sleep normally without being affected. Any interference.
  • the signal detection system for determining sleep apnea further includes an A/D conversion module, a buffer module, and a filter module that are sequentially connected to the signal output terminal of the vital sign signal collector;
  • the /D conversion module is used to convert the analog sign signal into a digital sign signal;
  • the input of the buffer module is the digital sign signal after A/D, and the output is the stack signal to be processed;
  • the filter module includes: low-pass filtering , Band-pass filter, morphological filter, three filter combinations, low-pass filter is used to remove high-frequency noise, band-pass filter is used to separate vital signs signals in each frequency band, and morphological filter is used to identify signal baseline values and low-frequency fluctuation characteristics. Remove signal baseline interference.
  • the vital sign signals of the user during sleep are acquired through the portable vital sign signal acquisition device, without the tester wearing electrodes, and the tester’s experience during the test is more comfortable; at the same time;
  • the vital sign signal is also structured to remove the signal noise, and the vital sign signal after the signal noise is removed is input into the signal detection model for signal processing to obtain the probability data of the occurrence of sleep apnea events, Thereby, it is convenient to improve the accuracy of judgment of apnea events.
  • the system uses the multi-dimensional feature set of the sleep breathing sample signal to perform feature training on the initial classification model, so that the performance of the trained classification model is more robust.
  • the system is expected to predict sleep apnea at home outside the hospital in the future. Screening pre-inspection has practical engineering reference significance, which is convenient for testers to obtain long-term measurements outside the hospital.

Abstract

一种用于判断睡眠呼吸暂停的信号检测方法及系统，信号检测方法包括以下步骤：获取用户睡眠时的生命体征信号（S1）；将用户睡眠时的生命体征信号进行结构化处理去除无效信号获得有效体征信号集合（S2）；提取并通过睡眠呼吸样本信号的多维形态特征对分类器初始模型进行特征训练获得睡眠呼吸检测模型（S3）；将有效体征信号集合输入到睡眠呼吸检测模型中进行信号处理获得用户睡眠时发生呼吸暂停的概率数据（S4）。用于判断睡眠呼吸暂停的信号检测方法，利用多维形态特征对分类器初始模型进行特征训练，使得睡眠呼吸检测模型的性能更加鲁棒，能够获得较为准确的用户睡眠时发生呼吸暂停的概率数据，从而便于判断睡眠呼吸暂停事件是否发生。

Description

一种用于判断睡眠呼吸暂停的信号检测方法及系统 技术领域
本发明涉及睡眠呼吸信号研究领域，特别是涉及一种用于判断睡眠呼吸暂停的信号检测方法及系统。
背景技术
睡眠是一种重要的生理活动，对于人体的物理和精神方面的自我恢复具有非常关键的作用。近年来，随着人们生活节奏的加快和工作压力的增加，人们对于自身的健康意识日益增强，各种便携式的医疗检测设备在家庭生活应用中得到了普及。而一般的便携式检测设备对获取的生命体征信号的处理方式较为简单，其分类方法以及特征工程检测方法多局限于单一维度的经验统计，未结合多维特征来对信号检测模型进行有效训练，因此，可能存在分类精准度不高，导致睡眠呼吸暂停事件判断不够准确的问题。
发明内容
为克服相关技术中存在的问题，本发明实施例提供了一种用于判断睡眠呼吸暂停的信号检测方法及系统。
根据本发明实施例的第一方面，提供一种用于判断睡眠呼吸暂停的信号检测方法，包括如下步骤：
获取用户睡眠时的生命体征信号；
将用户睡眠时的生命体征信号进行结构化处理去除无效信号获得有效体征信号集合；
提取并通过睡眠呼吸样本信号的多维形态特征对分类器初始模型进行特征训练获得睡眠呼吸检测模型；
将有效体征信号集合输入到睡眠呼吸检测模型中进行信号处理获得用户睡眠时发生呼吸暂停的概率数据。
本实施例所述的用于判断睡眠呼吸暂停的信号检测方法，利用多维形态特征对分类器初始模型进行特征训练以获得睡眠呼吸检测模型，使得睡眠呼吸检测模型的性能更加鲁棒，能够获得较为精准的用户睡眠时发生呼吸暂停的概率数据，从而便于用户或医生准确判断用户睡眠时是否发生呼吸暂停事件。
在一个可选的实施例中，将用户睡眠时的生命体征信号进行结构化处理去除无效信号，获得有效体征信号集合的方法包括如下步骤：
通过离床判决方法去除离床信号；
通过体动判决方法去除体动信号；
通过信号有效性判决去除无效信号区间；
将去除无效信号区间后的生命体征信号进行合理拼接，获得去除干扰后的有效体征信号集合。
在一个可选的实施例中，提取并通过睡眠呼吸样本信号的多维形态特征对分类器初始模型进行特征训练获得睡眠呼吸检测模型的方法，包括如下步骤：
对睡眠呼吸样本信号进行结构化处理去除无效信号，获取有效睡眠呼吸样本信号集合；
提取有效睡眠呼吸样本信号中的BCG样本信号；
提取固定时间尺度内的BCG样本信号的多维形态特征集合，所述多维形态特征包括：低频特征、峰值特征、面积特征、功率谱特征和非线性特征；
将提取的BCG样本信号的多维形态特征集合输入到集成学习模型中进行特征优化，获得稳态特征集合；
将稳态特征集合输入到多个分类器初始模型中进行特征分类训练获得睡眠呼吸检测模型。
在一个可选的实施例中，提取有效睡眠呼吸样本信号中的BCG样本信号的方法为：
识别BCG样本信号的J峰和K谷，定位固定时间尺度内的每个BCG样本信号的J峰位置和K谷位置；
将J峰向左遍历第一时间尺度，K峰向右遍历第二时间尺度，以定位出完整的BCG样本信号，并定位出固定时间尺度内的所有BCG样本信号；
筛选出固定时间尺度内的BCG样本信号。
在一个可选的实施例中，筛选出固定时间尺度内的BCG样本信号包括如下步骤：
计算固定时间尺度内的所有BCG样本信号的平均值，将该平均值作为BCG样本信号模型，所述BCG样本信号模型为：
Figure PCTCN2020128426-appb-000001
计算固定时间尺度内的所有BCG样本信号与BCG样本信号模型之间的归一化欧式距离以及归一化动态时间规整距离；
设定欧式距离预设阈值和动态时间规整距离预设阈值，将欧式距离大于欧式距离预设阈值，且归一化动态时间规整距离大于动态时间规整距离预设阈值的BCG信号丢弃，获得BCG样本信号。
在一个可选的实施例中，将提取的BCG样本信号的多维形态特征集合输入到集成学习模 型中进行特征优化的方法包括以下步骤：
将多维形态特征集合输入到树模型中进行样本特征训练，获得第一误差值；
对多维形态特征集合中的特定列进行随机上下置换，将特定列上下置换后的多维形态特征集合输入到树模型中进行样本特征训练，获得第二误差值；
计算出第一误差值和第二误差值的差值，并计算出差值的绝对值；
预设经验阈值，将第一误差值和第二误差值之间的差值的绝对值小于预设经验阈值的对应形态特征删除，获得优化后的特征集合；
将优化后的特征集合再次进行优化训练，获得稳态特征集合。
在一个可选的实施例中，所述分类器初始模型包括LR分类器、SVM分类器、RF分类器和AdaBoost分类器。
根据本发明实施例的第二方面，还提供了一种用于判断睡眠呼吸暂停的信号检测系统，包括：
生命体征信号采集装置，用于采集用户睡眠时的生命体征信号；
存储器，用于存储程序；
处理器，用于通过执行所述存储器存储的程序以实现如上任意一项所述的方法。
本发明实施例所述的用于判断睡眠呼吸暂停的信号检测系统，通过便携式，非接触的生命体征信号采集装置采集用户睡眠时的神秘感体征信号，给用户带来较好的测试体验，不会影响用户的正常睡眠，并且，该信号检测系统能够对用户睡眠时的生命体征信号进行噪声过滤，信号处理分析更加准确，以便于用户或医生对用户睡眠时发生呼吸暂停事件做出准确判断。
应当理解的是，以上的一般描述和后文的细节描述仅是示例性和解释性的，并不能限制本发明。
为了更好地理解和实施，下面结合附图详细说明本发明。
附图说明
图1为本发明实施例所述用于判断睡眠呼吸暂停的信号检测方法的流程示意图；
图2为本发明实施例所述用于判断睡眠呼吸暂停的信号检测方法的步骤2的流程示意图；
图3为本发明实施例所述用于判断睡眠呼吸暂停的信号检测方法的步骤3的流程示意图。
具体实施方式
这里将详细地对示例性实施例进行说明，其示例表示在附图中。下面的描述涉及附图时， 除非另有表示，不同附图中的相同数字表示相同或相似的要素。以下示例性实施例中所描述的实施方式并不代表与本发明相一致的所有实施方式。相反，它们仅是与如所附权利要求书中所详述的、本发明的一些方面相一致的装置和方法的例子。
在本发明使用的术语是仅仅出于描述特定实施例的目的，而非旨在限制本发明。在本发明和所附权利要求书中所使用的单数形式的“一种”、“所述”和“该”也旨在包括多数形式，除非上下文清楚地表示其他含义。还应当理解，本文中使用的术语“和/或”是指并包含一个或多个相关联的列出项目的任何或所有可能组合。
请参阅图1，其是本发明实施例所述用于判断睡眠呼吸暂停的信号检测方法的流程示意图。
本实施例所述用于判断睡眠呼吸暂停的信号检测方法包括如下步骤：
S1：获取用户睡眠时的生命体征信号；
S2：将用户睡眠时的生命体征信号进行结构化处理去除无效信号获得有效体征信号集合；
S3：提取并通过睡眠呼吸样本信号的多维形态特征对分类器初始模型进行特征训练获得睡眠呼吸检测模型；
S4：将有效体征信号集合输入到睡眠呼吸检测模型中进行信号处理获得用户睡眠时发生呼吸暂停的概率数据。
本发明实施例所述用于判断睡眠呼吸暂停的信号检测方法，通过提取睡眠呼吸样本信号的多维形态特征，利用多维形态特征对分类器初始模型进行特征训练以获得睡眠呼吸检测模型，并将结构化处理后的有效体征信号集合输入到睡眠呼吸检测模型中，进行信号处理以获得较为精准的用户睡眠时发生呼吸暂停的概率数据，从而便于用户或医生准确判断用户睡眠时是否发生呼吸暂停事件。
由于所采集到的连续时间内的信号中，会存在离床信号、体动信号、以及其他的无效信号等信号噪声，因此需要将上述无效信号去除，才能使得信号检测存在意义。
请参阅图2，其是本发明实施例所述用于判断睡眠呼吸暂停的信号检测方法的步骤2的流程示意图。
在步骤S2：将用户睡眠时的生命体征信号进行结构化处理去除无效信号获得有效体征信号集合的步骤中，还包括以下步骤：
S21：通过离床判决方法去除离床信号。
具体地，当用户处于离床状态时，所采集到的信号仅包括热噪声，为无效信号，需要去除。因此，所述离床判决方法为：可以根据高斯白噪声的统计特性，定义一个时间尺度T，对该时间尺度T内的睡眠呼吸信号进行分析，获取该时间尺度内的睡眠呼吸信号的数学期望、 功率谱密度、自相关和信号幅值；当数学期望近似为零，功率谱密度近似为常数，时间域信号自相关近似为冲击，且信号幅值小于预设固定阈值时，判决此时的睡眠呼吸信号为用户离床的信号，并将该信号去除。
S22：通过体动判决方法去除体动信号。
具体地，包括所述大体动信号去除方法和小体动信号去除方法。
所述大体动信号去除方法为：在固定时间尺度T内，根据信号的预设限定阈值确定大体动起止时间，并调节限幅阈值确定大体动信号，将大体动信号去除。
所述小体动信号去除方法为：采用希尔伯特变化计算睡眠呼吸信号中的包络函数mu(t)，并计算包络函数中最大值max{}与最小值min{}之比，以及最大值max{}与均值mean{}之比，当满足max{mu(t)}/min{mu(t)}>p1且max{mu(t)}/mean{mu(t)}>p2时，其中p1和p2分别为经验阈值，则判决为该时间尺度内存在小体动信号，调节限幅阈值确定小体动信号，并将小体动信号去除。
S23：通过信号有效性判决去除无效信号区间。
当用户身体与传感器距离过远，导致输出信号信噪比过低，不具备分析意义。该情况下，心冲击图与呼吸信号原本具备的近似周期性被噪声淹没，因此需要将无效信号区间去除。
具体地，所述信号有效性判决的方法为：通过经验模态分解或小波变换，分析固定时间尺度T内的生命体征信号不同频率区间的循环平稳特性与自相关的分析，确定该时间尺度内的睡眠呼吸信号是否有效，并将无效信号的对应区间去除。
S24：将去除无效信号区间后的生命体征信号进行合理拼接，获得去除干扰后的有效体征信号集合。
在一个实施例中，在睡眠呼吸样本信号进行合理拼接前还包括如下步骤：分别统计相邻区间内信号的一阶统计和二阶统计，当相邻区间内信号的一阶统计和二阶统计均小于预设固定阈值时，将数据直接合并拼接；当相邻区间内信号的一阶统计和二阶统计不小于预设固定阈值时，则将两段信号分别进行分类判决。
请参阅图3，其是本发明实施例所述用于判断睡眠呼吸暂停的信号检测方法的步骤3的流程示意图。
在步骤S3中，提取并通过睡眠呼吸样本信号的多维形态特征对分类器初始模型进行特征训练获得睡眠呼吸检测模型的方法，包括如下步骤：
S31：对睡眠呼吸样本信号进行结构化处理去除无效信号，获取有效睡眠呼吸样本信号集合；
S32：提取有效睡眠呼吸样本信号中的BCG样本信号；
S33：提取固定时间尺度内的BCG样本信号的多维形态特征集合，所述多维形态特征包括：低频特征、峰值特征、面积特征、功率谱特征和非线性特征；
S34：将提取的BCG样本信号的多维形态特征集合输入到集成学习模型中进行特征优化，获得稳态特征集合；
S35：将稳态特征集合输入到多个分类器初始模型中进行特征分类训练获得睡眠呼吸检测模型。
其中，步骤S32的实现方法为：
识别BCG样本信号的J峰和K谷，定位固定时间尺度内的每个BCG样本信号的J峰位置和K谷位置；
将J峰向左遍历第一时间尺度，K峰向右遍历第二时间尺度，以定位出完整的BCG样本信号，并定位出固定时间尺度内的所有BCG样本信号；
筛选出固定时间尺度内的BCG样本信号。
在一个实施例中，筛选出固定时间尺度内的BCG样本信号的方法为：
计算固定时间尺度内的所有BCG样本信号的平均值，将该平均值作为BCG样本信号模型，所述BCG样本信号模型为：
Figure PCTCN2020128426-appb-000002
计算固定时间尺度内的所有BCG样本信号与BCG样本信号模型之间的归一化欧式距离以及归一化动态时间规整距离；
设定欧式距离预设阈值和动态时间规整距离预设阈值，将欧式距离大于欧式距离预设阈值，且归一化动态时间规整距离大于动态时间规整距离预设阈值的BCG信号丢弃，获得BCG样本信号。
其中，在步骤S33中，多维形态特征的提取方法包括以下方法。
低频特征提取方法：
在固定时间尺度内，根据对BCG样本信号的J峰和K峰的识别，提取出J峰到K峰之间的信息，通过多次样条插值获取该时间窗内的上下包络函数，并对上下包络函数做经验模态分解以提取出上下包络函数中的低频分量，并定义为Eu(t)和Ed(t)。具体地，上下包络覆盖BCG信号J峰和K峰，将上包络定义为函数为mu(t)，下包络定义为函数为md(t)，并分别对mu(t)和md(t)做经验模态分解，提取上下包络函数中低频部分并定义为Eu(t)和Ed(t)。低频分量特征可真实反映出信号原本波动性，因而，Eu(t)和Ed(t)波动性和复杂度可作为判决呼吸暂停的特征之一。
峰值特征提取方法：
识别出固定时间尺度内的每个BCG样本信号的J峰以及K峰，构成集合{CJ(i)}与{CK(i)}，其中，i为当前时间尺度内BCG信号的个数。
由于BCG样本信号中J峰至K峰的形态特征最为鲁棒，且受到呼吸波起伏震荡效应，因此，还计算出集合{CJ(i)}与{CK(i)}的方差、标准差、峰度、斜度等，并将计算出的方差、标准差、峰度、斜度作为识别呼吸暂停的特征；
另外，还对{CJ(i)}与{CK(i)}集合中相邻数据的一阶差分进行计算，重新构造出新的集合{ΔCJ(i)}与{ΔCK(i)}，并进一步求解二阶差分，构造新的集合{Δ2CJ(i)}与{Δ2CK(i)}，将数据集合的一阶差分、二阶差分的集合进行一维、二维数据统计，计算出数值方差及标准差，将数值方差和标准差作为识别呼吸暂停判决的特征。
面积特征提取方法：
计算固定时间尺度内的每一个BCG样本信号的H峰、1谷、I峰、J谷、K峰以及L峰覆盖区域积分，即每一个BCG信号从H峰到L峰下的面积，并计算出BCG样本信号包络覆盖面积的方差和标准差，将方差和标准差作为识别呼吸暂停判决的特征。
功率谱特征提取方法：
对每一个完整BCG信号求解固定长度的傅里叶变换，量化相邻BCG信号功率谱密度的高低频之比，及固定时间尺度内所有BCG信号功率谱低频波动性，视为呼吸暂停判决的特征。
非线性特征提取方法：
将固定时间尺度下的原始信号降频处理，再进行样本熵的运算，获取样本熵。熵是随机变量不确定性的度量，不确定性越大，熵值越大。在本实施例中使用的是样本熵，样本熵在和其他的熵值计算上如近似熵，具有两个优势：首先，样本熵处理运算不需要考虑数据的长度；其次，样本熵具备了很好的一致性。经过样本熵运算的值越小，反应出序列本身的相似程度就越高；相反，样本熵运算的值越大，样本序列本身的就越紊乱、复杂。因此，样本熵在评估具有时间序列生理信号的紊乱程度和诊断病理方面均得到了很好的运用，以此来区分暂停片段和正常呼吸片段在数值上的差异。
其中，在步骤S34中：将提取的BCG样本信号的多维形态特征集合输入到集成学习模型中进行特征优化的方法，包括以下步骤：
将多维形态特征集合输入到树模型中进行样本特征训练，获得第一误差值，即对于任意一颗树模型Qi，i＝1,2,...，N，其中N为模型中树的个数，基于首次提取的特征样本进行训练，得到第一误差e1；
对多维形态特征集合中的特定列进行随机上下置换，将特定列上下置换后的多维形态特征集合输入到树模型中进行样本特征训练，获得第二误差值；具体地，随机改变特征集合中 的第j列，即特征j，保持其他列不变，对第j列进行随机的上下置换，得到第二误差e2；
计算出第一误差值和第二误差值的差值，并计算出差值的绝对值|e1-e2|；
预设经验阈值，将第一误差值和第二误差值之间的差值的绝对值小于预设经验阈值的对应形态特征删除，获得优化后的特征集合；即设定一个经验阈值β，当特征重要性|e1-e2|<β，则定义该特征对模型整体判决的贡献度有限或存在负向贡献，故删除该维度特征；
将优化后的特征集合再次进行优化训练，获得稳态特征集合。
在步骤S4中：将有效体征信号集合输入到睡眠呼吸检测模型中进行信号处理获得用户睡眠时发生呼吸暂停的-概率数据。具体地，分类判决的方法为：借助优化后的特征集合，采用N(N>1)种分类器对呼吸暂停进行分类，分类器包括LR、SVM、RF、AdaBoost等；结合不同分类器输出结果误差存在一定程度的统计独立性，将各分类器输出结果进行加权投票，即比值合并，最终输出针对呼吸暂停事件的二分类判决，从而便于发现长时间体征信号中非频发短暂异常片段。其中，加权系数与训练集误差系数呈反比例关系。
在一个可选的实施例中，为校验分类器分类结果的准确性，降低初步分类中误判漏检，还对所述概率数据进行基于不同时间尺度的定量分析，以判断所述概率数据的对应时间尺度内的片段信号是否存在波动性下降，若对应时间尺度内的片段信号存在波动性下降则说明该片段信号为睡眠呼吸暂停信号，可验证由睡眠呼吸检测模型所获得的概率数据较为准确。
基于不同时间尺度的定量分析的方法为：定义S＝{S 1,S 2,…,S N}为基于分类器初步检测的呼吸暂停信号片段，其中S n＝[s n(t),s n(t+1),…,s n(t+L)]，其中L为暂停时间长度，N为分类器检出暂停片段的数量，s n(t+l)为对应暂停片段在时间t+1时刻的信号。定义{T}为变长时间尺度，例如：T＝30s，60s，120s，计算在不同时间尺度{T}下暂停信号片段是否真实出现“波动性下降”的趋势，而具体检测方法采用基于多尺度归一化平均绝对离差分析方法或基于多尺度归一化方差分析方法。
其中，基于多尺度归一化平均绝对离差分析方法为：计算一定时间尺度下信号MAD(mean absolute error)值，定义该数值为M T；再此基础上以10s滑动窗上述时间尺度信号片段上滑动，计算10s时间窗滑动中的每一个MAD值，在对应时间尺度T下的MAD集合定义为M＝[M 1,M 2,…,M (T-10)]，计算c T＝min{M}/M T。由较小时间尺度分析获得的波动性为较小波动性，由较大时间尺度分析获得的波动性为较大波动性，而所述MAD值可以很好的描述信号的波动程度，在通过睡眠呼吸检测模型获得的概率数据之后，分析睡眠呼吸暂停信号片段在不同时间尺度下时的波动性，可反应出呼吸暂停信号片段相对正常呼吸片段是否存在显著波动性下降。当最小波动性与平均波动性之比小于0.2时，即当c T<0.2，则说明该呼吸暂停信号片段相对于正常呼吸片段存在显著波动性下降，由此可以验证分类的准确性。
另外，基于多尺度归一化方差分析方法也可以较好的描述信号稳态与交变的程度，通过基于多尺度归一化方差分析方法可以计算出多尺度下呼吸暂停信号片段与正常呼吸信号片段之间的归一化方差之比，也可以验证分类的准确性。
本发明实施例公开了一种用于判断睡眠呼吸暂停的信号检测方法，通过压电传感器采集生命体征信号，并对采集到的生命体征信号做结构化处理去除体动噪声以及其他噪声，然后将有效体征信号集合输入到基于多维形态特征训练出的睡眠呼吸检测模型中进行信号处理，获得用户睡眠时发生呼吸暂停的概率数据，以便于用户或医生更准确的判断用户睡眠中是否出现呼吸暂停事件，以及出现呼吸暂停事件的时间段。该方法对于未来院外居家睡眠呼吸暂停等预检具备实际工程参考意义。
本发明实施例还公开了一种用于判断睡眠呼吸暂停的信号检测系统，包括：生命体征信号采集装置，用于采集待检测者的生命体征信号；存储器，用于存储程序；处理器，用于通过执行所述存储器存储的程序以实现如上所述的方法。
所述生命体征信号采集装置为压电传感器模块，信号采集时，只需要将该压电传感器模块放置用户的头部下方即可长时间观测用户的睡眠呼吸情况，用户可以正常睡眠，不会受到任何干扰。
在一个实施例中，所述用于判断睡眠呼吸暂停的信号检测系统还包括依次连接于所述生命体征信号采集器的信号输出端的A/D转换模块、缓存模块和滤波模组；所述A/D转换模块用于将模拟体征信号转换成数字体征信号；所述缓存模块的输入为A/D后的数字体征信号，输出为待处理的堆栈信号；所述滤波模组包括：低通滤波、带通滤波、形态滤波三种滤波器组合，低通滤波用于去除高频噪声，带通滤波用于分离各个频段的生命体征信号，形态滤波用于识取信号基线数值及低频波动特性，去除信号基线干扰。
本发明实施例所述的用于判断睡眠呼吸暂停的信号检测，通过便携式生命体征信号采集装置获取用户睡眠时的生命体征信号，无需测试者佩戴电极，测试时给测试者的体验较为舒适；同时，在信号处理的过程中，还对生命体征信号进行结构化处理去除信号噪声，并将去除信号噪声后的生命体征信号输入到信号检测模型中进行信号处理获得睡眠呼吸暂停事件发生的概率数据，从而便于提高呼吸暂停事件判断的准确性。该在信号处理的过程中，本系统通过睡眠呼吸样本信号的多维特征集合对分类初始模型进行特征训练，从而使得训练出的分类模型性能更加鲁棒，该系统对于未来院外居家睡眠呼吸暂停的预筛预检具备实际工程参考意义，便于测试者在院外获得长时间的测量，十分方便。
以上所述实施例仅表达了本发明的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不 脱离本发明构思的前提下，还可以做出若干变形和改进，这些都属于本发明的保护范围。

Claims (8)

  1. 一种用于判断睡眠呼吸暂停的信号检测方法，其特征在于，包括如下步骤：
    获取用户睡眠时的生命体征信号；
    将用户睡眠时的生命体征信号进行结构化处理去除无效信号获得有效体征信号集合；
    提取并通过睡眠呼吸样本信号的多维形态特征对分类器初始模型进行特征训练获得睡眠呼吸检测模型；
    将有效体征信号集合输入到睡眠呼吸检测模型中进行信号处理获得用户睡眠时发生呼吸暂停的概率数据。
  2. 根据权利要求1所述用于判断睡眠呼吸暂停的信号检测方法，其特征在于，
    将用户睡眠时的生命体征信号进行结构化处理去除无效信号，获得有效体征信号集合的方法包括如下步骤：
    通过离床判决方法去除离床信号；
    通过体动判决方法去除体动信号；
    通过信号有效性判决去除无效信号区间；
    将去除无效信号区间后的生命体征信号进行合理拼接，获得去除干扰后的有效体征信号集合。
  3. 根据权利要求1所述用于判断睡眠呼吸暂停的信号检测方法，其特征在于，
    提取并通过睡眠呼吸样本信号的多维形态特征对分类器初始模型进行特征训练获得睡眠呼吸检测模型的方法，包括如下步骤：
    对睡眠呼吸样本信号进行结构化处理去除无效信号，获取有效睡眠呼吸样本信号集合；
    提取有效睡眠呼吸样本信号中的BCG样本信号；
    提取固定时间尺度内的BCG样本信号的多维形态特征集合，所述多维形态特征包括：低频特征、峰值特征、面积特征、功率谱特征和非线性特征；
    将提取的BCG样本信号的多维形态特征集合输入到集成学习模型中进行特征优化，获得稳态特征集合；
    将稳态特征集合输入到多个分类器初始模型中进行特征分类训练获得睡眠呼吸检测模型。
  4. 根据权利要求3所述用于判断睡眠呼吸暂停的信号检测方法，其特征在于，
    提取有效睡眠呼吸样本信号中的BCG样本信号的方法为：
    识别BCG样本信号的J峰和K谷，定位固定时间尺度内的每个BCG样本信号的J峰位 置和K谷位置；
    将J峰向左遍历第一时间尺度，K峰向右遍历第二时间尺度，以定位出完整的BCG样本信号，并定位出固定时间尺度内的所有BCG样本信号；
    筛选出固定时间尺度内的BCG样本信号。
  5. 根据权利要求4所述用于判断睡眠呼吸暂停的信号检测方法，其特征在于，筛选出固定时间尺度内的BCG样本信号包括如下步骤：
    计算固定时间尺度内的所有BCG样本信号的平均值，将该平均值作为BCG样本信号模型，所述BCG样本信号模型为：
    Figure PCTCN2020128426-appb-100001
    计算固定时间尺度内的所有BCG样本信号与BCG样本信号模型之间的归一化欧式距离以及归一化动态时间规整距离；
    设定欧式距离预设阈值和动态时间规整距离预设阈值，将欧式距离大于欧式距离预设阈值，且归一化动态时间规整距离大于动态时间规整距离预设阈值的BCG信号丢弃，获得BCG样本信号。
  6. 根据权利要求3所述用于判断睡眠呼吸暂停的信号检测方法，其特征在于，
    将提取的BCG样本信号的多维形态特征集合输入到集成学习模型中进行特征优化的方法包括以下步骤：
    将多维形态特征集合输入到树模型中进行样本特征训练，获得第一误差值；
    对多维形态特征集合中的特定列进行随机上下置换，将特定列上下置换后的多维形态特征集合输入到树模型中进行样本特征训练，获得第二误差值；
    计算出第一误差值和第二误差值的差值，并计算出差值的绝对值；
    预设经验阈值，将第一误差值和第二误差值之间的差值的绝对值小于预设经验阈值的对应形态特征删除，获得优化后的特征集合；
    将优化后的特征集合再次进行优化训练，获得稳态特征集合。
  7. 根据权利要求3所述用于判断睡眠呼吸暂停的信号检测方法，其特征在于，
    所述分类器初始模型包括LR分类器、SVM分类器、RF分类器和AdaBoost分类器。
  8. 一种用于判断睡眠呼吸暂停的信号检测系统，其特征在于，包括：
    生命体征信号采集装置，用于采集用户睡眠时的生命体征信号；
    存储器，用于存储程序；
    处理器，用于通过执行所述存储器存储的程序以实现如权利要求1-7中任意一项所述的方法。
PCT/CN2020/128426 2019-12-31 2020-11-12 一种用于判断睡眠呼吸暂停的信号检测方法及系统 WO2021135672A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US17/790,445 US20230043406A1 (en) 2019-12-31 2020-11-12 Noninvasive method and system for sleep apnea detection
US18/480,906 US20240023886A1 (en) 2019-12-31 2023-10-04 Noninvasive method and system for sleep apnea detection

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911413560.9 2019-12-31
CN201911413560.9A CN111035367B (zh) 2019-12-31 2019-12-31 一种用于判断睡眠呼吸暂停的信号检测系统

Related Child Applications (2)

Application Number Title Priority Date Filing Date
US17/790,445 A-371-Of-International US20230043406A1 (en) 2019-12-31 2020-11-12 Noninvasive method and system for sleep apnea detection
US18/480,906 Continuation US20240023886A1 (en) 2019-12-31 2023-10-04 Noninvasive method and system for sleep apnea detection

Publications (1)

Publication Number Publication Date
WO2021135672A1 true WO2021135672A1 (zh) 2021-07-08

Family

ID=70242458

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/128426 WO2021135672A1 (zh) 2019-12-31 2020-11-12 一种用于判断睡眠呼吸暂停的信号检测方法及系统

Country Status (3)

Country Link
US (2) US20230043406A1 (zh)
CN (1) CN111035367B (zh)
WO (1) WO2021135672A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113633280A (zh) * 2021-08-13 2021-11-12 上海跃扬医疗科技有限公司 在床离床检测方法、检测系统、设备及存储介质
CN117017233A (zh) * 2023-10-08 2023-11-10 华南师范大学 一种睡眠呼吸暂停检测方法、装置、设备以及存储介质

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111035367B (zh) * 2019-12-31 2021-05-18 华南师范大学 一种用于判断睡眠呼吸暂停的信号检测系统
CN111814830A (zh) * 2020-06-08 2020-10-23 珠海格力电器股份有限公司 睡眠状态检测模型构建、睡眠状态检测方法及装置
CN111938588A (zh) * 2020-07-24 2020-11-17 深圳数联天下智能科技有限公司 一种检测睡眠状态的方法、睡眠监测仪及存储介质
US11925447B1 (en) * 2021-02-24 2024-03-12 Amazon Technologies, Inc. Out-of-bed zone generation and monitoring associated with respiratory activity tracking
CN114010185A (zh) * 2021-11-17 2022-02-08 上海跃扬医疗科技有限公司 一种呼吸障碍的监测方法、装置、设备及介质
CN113951869A (zh) * 2021-11-17 2022-01-21 上海跃扬医疗科技有限公司 一种呼吸障碍检测方法、装置、设备及介质
CN114027825B (zh) * 2022-01-10 2022-03-22 华南师范大学 一种呼吸信号获取方法、装置和计算机设备
CN114668407B (zh) * 2022-05-30 2023-01-10 广州中科新知科技有限公司 一种基于bcg信号的睡眠呼吸暂停初筛系统
CN114767127B (zh) * 2022-06-21 2022-09-30 广州中科新知科技有限公司 一种心冲击图信号的处理方法及系统
CN115486833B (zh) * 2022-08-22 2023-06-06 华南师范大学 呼吸状态检测方法、装置、计算机设备以及存储介质
CN115381438B (zh) * 2022-08-24 2023-04-18 华南师范大学 生命体征信号的重建方法、装置、计算机设备和存储介质
CN116095922B (zh) * 2023-03-13 2023-08-18 广州易而达科技股份有限公司 一种照明灯控制方法、装置、照明灯和存储介质
CN116369907B (zh) * 2023-03-20 2024-02-13 华南师范大学 一种心冲击图信号定位方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030055348A1 (en) * 2001-09-14 2003-03-20 University College Dublin Apparatus for detecting sleep apnea using electrocardiogram signals
CN202568219U (zh) * 2012-03-01 2012-12-05 北京麦邦光电仪器有限公司 睡眠心率、呼吸监测系统
CN106037671A (zh) * 2016-07-11 2016-10-26 西北工业大学 一种基于bcg信号的呼吸暂停事件检测方法及系统
CN107361770A (zh) * 2017-08-15 2017-11-21 杭州医电园科技有限公司 睡眠呼吸暂停事件判别装置
CN109394188A (zh) * 2018-11-27 2019-03-01 中山大学 一种基于心率变异性的呼吸异常检测方法、装置及设备
CN111035367A (zh) * 2019-12-31 2020-04-21 华南师范大学 一种用于判断睡眠呼吸暂停的信号检测方法及系统

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2011009085A2 (en) * 2009-07-17 2011-01-20 Oregon Health & Science University Method and apparatus for assessment of sleep disorders
JP5515156B2 (ja) * 2012-03-30 2014-06-11 株式会社タニタ 睡眠管理システム及び睡眠計
US10874577B2 (en) * 2013-03-15 2020-12-29 Somne Llc Obtaining, with a sleep-apnea device, information related to sleep-apnea events and sleep-apnea treatment, and correlating sleep apnea events and sleep-apnea treatment with subject lifestyle and wellbeing
CN104970772A (zh) * 2015-07-20 2015-10-14 翁南帮 人在静卧状态时生理参数的采集、呼吸暂停唤醒的方法及装置
FI126600B (en) * 2015-08-10 2017-03-15 Murata Manufacturing Co Detection of sleep phenomena using ballistocardiography
CN105559754B (zh) * 2015-12-29 2019-02-26 西北工业大学 一种基于心率与呼吸信号的睡眠呼吸障碍检测方法和装置
US20190314192A1 (en) * 2016-10-31 2019-10-17 Mc10, Inc. Closed Loop Respiratory Monitoring System For Sleep Quality Characterization
US10617311B2 (en) * 2017-08-09 2020-04-14 Samsung Electronics Co., Ltd. System and method for real-time heartbeat events detection using low-power motion sensor
EP3829435A1 (en) * 2018-07-27 2021-06-09 Murata Manufacturing Co., Ltd. Providing temporal information of a subject
CN110477887A (zh) * 2019-09-18 2019-11-22 浙江理工大学 一种非入侵式长程呼吸暂停综合征的监测装置

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030055348A1 (en) * 2001-09-14 2003-03-20 University College Dublin Apparatus for detecting sleep apnea using electrocardiogram signals
CN202568219U (zh) * 2012-03-01 2012-12-05 北京麦邦光电仪器有限公司 睡眠心率、呼吸监测系统
CN106037671A (zh) * 2016-07-11 2016-10-26 西北工业大学 一种基于bcg信号的呼吸暂停事件检测方法及系统
CN107361770A (zh) * 2017-08-15 2017-11-21 杭州医电园科技有限公司 睡眠呼吸暂停事件判别装置
CN109394188A (zh) * 2018-11-27 2019-03-01 中山大学 一种基于心率变异性的呼吸异常检测方法、装置及设备
CN111035367A (zh) * 2019-12-31 2020-04-21 华南师范大学 一种用于判断睡眠呼吸暂停的信号检测方法及系统

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113633280A (zh) * 2021-08-13 2021-11-12 上海跃扬医疗科技有限公司 在床离床检测方法、检测系统、设备及存储介质
CN113633280B (zh) * 2021-08-13 2023-08-22 上海跃扬医疗科技有限公司 在床离床检测方法、检测系统、设备及存储介质
CN117017233A (zh) * 2023-10-08 2023-11-10 华南师范大学 一种睡眠呼吸暂停检测方法、装置、设备以及存储介质
CN117017233B (zh) * 2023-10-08 2024-01-09 华南师范大学 一种睡眠呼吸暂停检测方法、装置、设备以及存储介质

Also Published As

Publication number Publication date
CN111035367B (zh) 2021-05-18
US20230043406A1 (en) 2023-02-09
CN111035367A (zh) 2020-04-21
US20240023886A1 (en) 2024-01-25

Similar Documents

Publication Publication Date Title
WO2021135672A1 (zh) 一种用于判断睡眠呼吸暂停的信号检测方法及系统
CN108670200B (zh) 一种基于深度学习的睡眠鼾声分类检测方法及系统
CN110135285B (zh) 一种使用单导设备的脑电静息态身份认证方法及装置
Sarno et al. Detection of diabetes from gas analysis of human breath using e-Nose
CN114010171B (zh) 一种基于心跳数据的分类器设置方法
Lu et al. Classification of single-channel EEG signals for epileptic seizures detection based on hybrid features
Hamza et al. Svm for human identification using the ECG signal
CN112869716B (zh) 一种基于双通道卷积神经网络的脉搏特征识别系统及方法
Gürkan et al. A novel biometric authentication approach using electrocardiogram signals
CN111370124A (zh) 一种基于面手部识别和大数据的健康分析系统及方法
CN111783715A (zh) 基于脉搏信号特征提取的身份识别方法
CN108962379B (zh) 一种脑神经系统疾病的手机辅助检测系统
Porieva et al. Investigation of lung sounds features for detection of bronchitis and COPD using machine learning methods
Gurkan et al. A novel human identification system based on electrocardiogram features
CN116616709A (zh) 基于非静止状态下多通道视频的工作记忆负荷评估方法
CN116451110A (zh) 基于信号能量特征和脉搏周期的血糖预测模型构建方法
CN113273992B (zh) 信号处理方法及装置
CN114550932A (zh) 一种睡眠呼吸暂停风险评估方法、装置、设备及存储介质
CN106344008B (zh) 睡眠状态分析中清醒状态检测方法和系统
CN114869272A (zh) 姿势震颤检测模型、姿势震颤检测算法、以及姿势震颤检测设备
CN114469041A (zh) 一种运动过程中心率变化数据特征分析方法
CN112890830A (zh) 一种基于睡眠脑网络的抑郁症患者数据分类方法及装置
CN113080966A (zh) 一种基于睡眠分期的抑郁症自动检测方法
CN112507959A (zh) 一种基于视频中个体面部分析的情绪感知模型的建立方法
CN111012306A (zh) 基于双神经网络的睡眠呼吸声检测方法及系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20909060

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20909060

Country of ref document: EP

Kind code of ref document: A1