CN111345798A - Method for detecting atrial fibrillation - Google Patents

Method for detecting atrial fibrillation Download PDF

Info

Publication number
CN111345798A
CN111345798A CN201811559121.4A CN201811559121A CN111345798A CN 111345798 A CN111345798 A CN 111345798A CN 201811559121 A CN201811559121 A CN 201811559121A CN 111345798 A CN111345798 A CN 111345798A
Authority
CN
China
Prior art keywords
sample entropy
sequence
time series
atrial fibrillation
time
Prior art date
Legal status (The legal status 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 status listed.)
Pending
Application number
CN201811559121.4A
Other languages
Chinese (zh)
Inventor
刘至伟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Youfang Information Technology Shanghai Co ltd
Original Assignee
Youfang Information Technology Shanghai Co ltd
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 Youfang Information Technology Shanghai Co ltd filed Critical Youfang Information Technology Shanghai Co ltd
Priority to CN201811559121.4A priority Critical patent/CN111345798A/en
Publication of CN111345798A publication Critical patent/CN111345798A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/024—Detecting, measuring or recording pulse rate or heart rate
    • 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
    • A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Veterinary Medicine (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • Public Health (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Physiology (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Psychiatry (AREA)
  • Cardiology (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Abstract

A method of detecting atrial fibrillation is disclosed. The method comprises the steps of obtaining a time sequence of heartbeat intervals with a certain time length; converting the acquired time series of heartbeat intervals into a de-trended time series; calculating a first sample entropy of the detrended time series and distinguishing atrial fibrillation from sinus rhythm by the first sample entropy; a second sample entropy of the detrended time series is calculated and the arrhythmia other than atrial fibrillation and sinus rhythm is discriminated with the second sample entropy. The method more effectively distinguishes atrial fibrillation from sinus rhythm and other arrhythmia fluctuations by using a time sequence of heart beat intervals of lesser length.

Description

Method for detecting atrial fibrillation
Technical Field
The present application relates to the field of physiological information detection, and more particularly, to a method for detecting atrial fibrillation.
Background
Atrial fibrillation (AFib) is a common chronic cardiac disorder, often accompanied by the risk of ischemic stroke. Patients with this disorder are often asymptomatic and fail to find early. Nowadays, in order to enable patients to find the disease earlier, Electrocardiogram (ECG), heartbeat pulse or blood vessel pulsation detecting devices are used to assist in detecting atrial fibrillation (AFib). The device for detecting heartbeat includes related devices using Photoplethysmography (PPG), such as a blood oxygen monitor or a wrist watch. The blood vessel pulsation detecting device is, for example, a blood pressure monitor.
Atrial fibrillation (AFib) is the result of abnormal electrical discharges in the Atrial Activity (AA), while normal heart rhythm is regulated by electrical signals from the sinoatrial node in the atrium. When the atrium discharges abnormally, on one hand, a continuous high-frequency irregular atrial activity electric signal can be detected on the atrium, and on the other hand, an original heart rhythm regulation mechanism is abnormal to generate random beats, so that an inter-beat interval (IBI) sequence with a certain time length changes and has the characteristics of large fluctuation variation and irregularity. Therefore, there are two ways to detect atrial fibrillation (AFib): detecting atrial activity electrical signals obtained by an Electrocardiogram (ECG); second, the heart beat interval (IBI) sequence is obtained by R-R wave interval (RRI) sequence of Electrocardiogram (ECG), optical signal of photoplethysmography (PPG) or peak-to-peak interval (PPI) sequence of pulse signal of hemomanometer.
In the first method, since only Electrocardiogram (ECG) signals are used for analysis, and the energy of the Atrial Activity (AA) waveform in the ECG waveform is small, if an ECG waveform with good signal quality and accuracy is to be obtained, the hardware, filtering capability, electrode material, electrode position, etc. of the ECG device must have high specifications, and even a certain time is required for the detection process to accumulate the spectral energy, so that proper analysis and accurate prediction can be achieved. Therefore, the detection method is not highly likely to be applied to the wearable device.
In the second method, the heartbeat interval (IBI) sequence can be obtained by an Electrocardiogram (ECG) device, a photoplethysmography (PPG) application device, an electronic sphygmomanometer device, or any physiological signal device that can obtain heartbeat time information, i.e., can be detected as long as a heartbeat rhythm or pulse time signal is obtained. Therefore, the wearing or mounting possibility is high. However, the main difficulty of this method is that both the variability and irregularity of the heartbeat fluctuation are observed through a certain amount of statistics to distinguish atrial fibrillation (AFib) from other heart rhythm problems. Other rhythm problems that are common are sinus rhythm (NSR) and arrhythmic rhythm (arrythmia, ARR), the fluctuation of the sequence of beat intervals (IBI) of sinus rhythm (NSR) also being irregular, but usually being less variable than atrial fibrillation (AFib), and the fluctuation of the sequence of beat intervals (IBI) of arrhythmic rhythm (ARR) being more variable but often in a particular arrangement (pattern). In other words, this method requires a heartbeat interval (IBI) sequence, R-R wave interval (RRI) sequence, or peak-to-peak interval (PPI) sequence of a certain length of time. In addition, since the heart beat interval (IBI) sequence is not the first-hand information of atrial fibrillation (AFib), but the heart rhythm regulation result generated by atrial arrhythmia caused by atrial fibrillation (AFib), the cumulative heart beat interval (IBI) sequence can be sampled only from the heart beat time of each heart beat.
Based on the second method, there is a document 1 (Tareno K, Glass L, "Automatic detection of the coefficient of variation and sensitivity of the historical results RR and Δ RR intermediates," Medical and Biological Engineering and monitoring 2001; 39(6): 664-671) proposes that the Standard Deviation (SD) of the "RRI sequence" of atrial fibrillation (AFib) and the "sequence formed by the difference (delta RR, Δ RR) of the adjacent RRI in the RRI sequence" is a certain proportion of the average (mean RR) "of the" RRI sequence. Wherein the standard deviation of the R-R wave spacing (RRI) sequence for atrial fibrillation (AFib) is 0.24 times the mean of the R-R wave spacing (RRI) sequence, and the standard deviation of the sequence formed by the difference of adjacent R-R wave Spacings (RRIs) for atrial fibrillation (AFib) is 0.34 times the mean of the R-R wave spacing (RRI) sequence. However, the statistical method of document 1 uses first-order statistics excluding temporal (sequential) characteristics, and thus can only roughly distinguish atrial fibrillation (AFib) from sinus rhythm (NSR) having a variation smaller than that of atrial fibrillation, and cannot effectively distinguish atrial fibrillation (AFib) from Arrhythmia (ARR) having a wave with a special arrangement.
Based on the second approach, another document 2(s. hardigitai, "Is indicative to detect atrial fibrillation by simple rhythm using RR intervals. Document 2 suggests that since fluctuations in the beat interval (IBI) sequence of atrial fibrillation (AFib) have no specific arrangement and are high in random irregularity compared with other Arrhythmia (ARR), the randomness is evaluated by the sample entropy, and thus atrial fibrillation (AFib) has a high randomness of sample entropy compared with other Arrhythmia (ARR). On the other hand, although sinus rhythm (NSR) has some irregularity and randomness, as long as the parameter-tolerance (tolerance) γ for evaluating the sample entropy is much higher than the fluctuation of sinus rhythm (NSR), the randomness of the beat of sinus rhythm (NSR) is neglected, so that atrial fibrillation (AFib) still has a high randomness of sample entropy compared with sinus rhythm (NSR). The method of document 2 is as follows: using a sequence of 50R-R wave pitches (RRIs), the dimension (dimension) m of the entropy of the evaluation sample is set to 2, and the tolerance γ of the entropy of the evaluation sample is set to a certain proportion of the mean value of the sequence of R-R wave pitches (RRIs), unlike the conventional method in which the tolerance γ of the entropy of the evaluation sample is set to a certain proportion of the standard deviation of the sequence of R-R wave pitches (RRIs).
Since the variability of atrial fibrillation (AFib) tends to be larger than that of sinus rhythm (NSR), i.e., the standard deviation of the R-R wave interval (RRI) sequence of atrial fibrillation (AFib) tends to be higher than that of the R-R wave interval (RRI) sequence of sinus rhythm (NSR), the tolerance γ of the evaluation sample entropy is set to a certain proportion of the mean value of the R-R wave interval (RRI) sequence, and for atrial fibrillation (AFib), the tolerance γ of the evaluation sample entropy is obtained as an appropriate proportion of the standard deviation of the R-R wave interval (RRI) sequence of atrial fibrillation (AFib) as a certain proportion of the mean value of the R-R wave interval (RRI) sequence, which is still equivalent to the conventional tolerance γ of the evaluation sample entropy as the standard deviation of the R-R wave interval (RRI) sequence, but for sinus rhythm (NSR), the tolerance γ of the evaluation sample entropy is adopted as compared to the conventional tolerance γ of the R-R wave interval (RRI) sequence The tolerance γ of the entropy of the evaluated samples is significantly higher, so that the small variations of the randomness can be neglected. Therefore, the method of document 2 can effectively distinguish atrial fibrillation (AFib) from sinus rhythm (NSR) using overestimated γ, as well as atrial fibrillation (AFib) from other Arrhythmia (ARR) with sample entropy.
However, the method of document 2 still has the following problems: 1. by adopting an R-R wave interval (RRI) sequence and setting the dimension m of the evaluation sample entropy to be 2, when the sinus rhythm (NSR) has a large variation tendency, the difference between measurement levels of the R-R wave interval (RRI) sequence at different time points can be also included in the sample entropy evaluation, so that the sample entropy evaluation value of the sinus rhythm (NSR) cannot be effectively reduced, and the partial situation is difficult to distinguish from the atrial fibrillation (AFib). 2. For other Arrhythmia (ARR), the sample entropy is evaluated with a certain proportion of the average value of the R-R wave interval (RRI) sequence as the tolerance γ of the evaluation sample entropy, i.e. equivalent to the tolerance γ of the order of atrial fibrillation (AFib), and for some Arrhythmia (ARR), the tolerance of the evaluation sample entropy of the obtained R-R wave interval (RRI) sequence is underestimated, which may not accurately reflect the low sample entropy characteristic of Arrhythmia (ARR), making Arrhythmia (ARR) difficult to distinguish from atrial fibrillation (AFib).
Therefore, how to overcome the above problems, it is a technical problem to be solved by the present application to develop a simple method for detecting atrial fibrillation (af) without too many heartbeat interval (IBI) sequences or R-R wave interval (RRI) sequences and effectively distinguishing atrial fibrillation (AFib), sinus rhythm (NSR) and arrhythmic rhythm (ARR), and apply it to any physiological signal device capable of obtaining heartbeat time information, including a single lead push type Electrocardiogram (ECG) device, a reflection type photoplethysmography (PPG) application device, a blood pressure machine, etc., to assist a patient to find atrial fibrillation (AFib) at an early stage.
Disclosure of Invention
In view of the foregoing, the present application provides a method for detecting atrial fibrillation.
In one embodiment, the present application provides a method for detecting atrial fibrillation, comprising the steps of: obtaining a time sequence of heartbeat intervals of a certain time length; converting the acquired time series of heartbeat intervals of a certain time length into a de-trended time series; taking the product of the average value of the acquired time series of the heartbeat intervals with a certain time length and a first proportion as the tolerance of a first sample entropy of the calculation detrended time series under a sample entropy evaluation dimension, and distinguishing atrial fibrillation from sinus rhythm by the first sample entropy; and calculating a tolerance of a second sample entropy of the detrended time series in the same sample entropy evaluation dimension by taking a product of a standard deviation of the detrended time series and a second proportion, and distinguishing arrhythmia except atrial fibrillation and sinus rhythm by the second sample entropy; wherein the time length of the time series of heartbeat intervals is less than 30 seconds, the tolerance for computing the entropy of the second sample is different from the tolerance for computing the entropy of the first sample, and the evaluation dimension of the entropy of the sample is at most 3.
In one embodiment, the first ratio is 0.034 to 0.085 in the proposed method for detecting atrial fibrillation.
In one embodiment, the second ratio is 0.1 to 0.25 in the proposed method for detecting atrial fibrillation. Optionally, the second ratio is 0.2.
In one embodiment, the method for detecting atrial fibrillation obtains a time sequence of heartbeat intervals of a certain time length as an R-R wave interval sequence, the detrended time sequence is a sequence formed by differences between adjacent R-R wave intervals of the R-R wave interval sequence, and the sample entropy evaluation dimension is 1.
In one embodiment, the step of converting the time series of heart beat intervals of a certain time length into a detrended time series further comprises filtering out heart beat fluctuations with a frequency of less than 0.4 HZ.
In another embodiment, the present application provides a method for detecting atrial fibrillation, comprising the steps of: obtaining a time sequence of heartbeat intervals of a certain time length; converting the acquired time series of heartbeat intervals of a certain time length into a de-trended time series; calculating a first sample entropy of the detrended time series, and distinguishing atrial fibrillation from sinus rhythm by the first sample entropy; and calculating a second sample entropy of the detrended time series, and discriminating cardiac arrhythmias other than atrial fibrillation and sinus rhythm with the second sample entropy; wherein the tolerance for calculating the second sample entropy is different from the tolerance for calculating the first sample entropy, the tolerance for calculating the first sample entropy is higher than the tolerance for calculating the sample entropy of the time series of heartbeat intervals of a certain length of time, and the tolerance for calculating the second sample entropy is lower than the tolerance for calculating the sample entropy of the time series of heartbeat intervals of a certain length of time.
In another embodiment, the method for detecting atrial fibrillation further comprises filtering out beat fluctuations with a frequency of less than 0.4HZ in the step of converting the time series of beat intervals of a certain length of time into the detrended time series.
In another embodiment, a method for detecting atrial fibrillation is provided in which a time series of heartbeat intervals of a certain length is obtained as an R-R wave interval series, and a detrended time series is a series of differences between adjacent R-R wave intervals of the R-R wave interval series.
In another embodiment, the time duration of the time series of heartbeat intervals of a certain time duration is less than 30 seconds.
According to the method for detecting atrial fibrillation provided by the embodiments of the present application, before detecting atrial fibrillation by using a temporal heartbeat interval (IBI) sequence, the temporal heartbeat interval (IBI) sequence is converted into a de-trended temporal sequence. The length of time for a sequence of so-called heartbeat intervals (IBIs) may be less than the length of time typically required to detect atrial fibrillation. So-called detrending also includes methods for deriving high frequency fluctuations in the sequence of heart beat intervals by low pass filtering, median filtering, high pass filtering or other similar means. Furthermore, for detrended time series, two different Sample entropies (Sample entrypy, SampEn) are used to distinguish atrial fibrillation (AFib), sinus rhythm (NSR) and other Arrhythmia (ARR), wherein one calculated Sample Entropy is more tolerant than the Sample Entropy calculated from the acquired time series of heart beat intervals of a certain time length and the other calculated Sample Entropy is less tolerant than the Sample Entropy calculated from the acquired time series of heart beat intervals of a certain time length, in such a way that in case of a large trend change of sinus rhythm (NSR), the difference between the magnitudes of RRI series at different time points is taken into the Sample Entropy calculation (SampEn) and Arrythmia (ARR) which may be misjudged as atrial fibrillation (AFib) is excluded. Thus, the proposed method can effectively distinguish between atrial fibrillation (AFib), sinus rhythm (NSR) and other Arrhythmias (ARR) without requiring a sequence of most heart beat intervals.
Drawings
FIG. 1 is a flow chart illustrating steps in a method for detecting atrial fibrillation according to an embodiment of the present application.
FIG. 2 is a flow chart illustrating steps in a method for detecting atrial fibrillation according to an embodiment of the present disclosure.
FIG. 3 is a flow chart illustrating steps in a method for detecting atrial fibrillation according to another embodiment of the present application.
FIG. 4A is a graph of a trending R-R wave interval (RRI) sequence of sinus rhythm (NSR) using the method of detecting atrial fibrillation according to the embodiments of the present disclosure.
FIG. 4B is a sequence diagram formed by the difference of R-R wave Spacings (RRIs) in the R-R wave spacing (RRI) sequence of the trending sinus rhythm (NSR) of FIG. 4A.
FIG. 5 is a sequence diagram of the R-R wave spacing (RRI) of other Arrhythmia (ARR) with a certain arrangement of fluctuations and not completely random, using the method of detecting atrial fibrillation according to the embodiments of the present application.
Detailed Description
The methods of atrial fibrillation, the causes involved, and known methods and principles of detection are well known to those of ordinary skill in the art and, therefore, are not described in full below. Also, the drawings referred to below are intended to convey meanings related to features of the present application, and are not necessarily drawn to scale.
FIG. 1 is a flow chart illustrating steps in a method for detecting atrial fibrillation according to an embodiment of the present application. As shown in fig. 1, a method for detecting atrial fibrillation according to one embodiment of the present application includes the following steps.
Step 11: a time series of heartbeat intervals (IBIs) of a certain length of time is taken. Next, step 12 is performed. Atrial fibrillation (AFib) is detected based on a heartbeat interval (IBI) sequence obtained from a R-R wave interval (RRI) sequence of an Electrocardiogram (ECG) or a peak-to-peak interval (PPI) sequence in a photoplethysmography (PPG) or a pulse signal of a sphygmomanometer, and thus, a heartbeat interval (IBI) sequence, an R-R wave interval (RRI) sequence or a peak-to-peak interval (PPI) sequence of a certain length of time is required. In one embodiment, this step is implemented by obtaining a sequence of R-R wave Spacings (RRIs) over a period of time. In addition, the time length of the obtained R-R wave interval (RRI) sequence with a certain time length can be less than 30 seconds, and the number of samples corresponding to the required R-R wave interval (RRI) sequence can be less than 50, so that the method does not need a large amount of data statistics to observe the variation degree and the irregularity characteristic of heartbeat fluctuation, is more easily applied to a wearable device, and improves the detection efficiency.
Step 12, convert the time sequence of heartbeat intervals of a certain time length into a time sequence of detrending (detrended), then, execute step 13. in order to solve the problem that the difference between the magnitudes of RRI sequences at different time points is taken into the sample entropy (SampEn) evaluation in the case of heart rhythm with large trend change, the time sequence of heartbeat intervals of a certain time length is converted into a time sequence of detrending before detecting atrial fibrillation by using the time sequence of heartbeat intervals (IBI), the method of obtaining high frequency fluctuation in the heartbeat interval sequence by low pass filtering, median filtering, high pass filtering or other similar means is included.
Step 13: a first sample entropy of the detrended time series is calculated (SampEn1) and atrial fibrillation (AFib) and sinus rhythm (NSR) are distinguished by the first sample entropy (SampEn 1). Next, step 14 is performed.
Step 14: a second sample entropy of the detrended time series is calculated (SampEn2) and atrial fibrillation (AFib) and arrhythmic Arrhythmia (ARR) other than sinus rhythm are distinguished by the second sample entropy (SampEn 2).
In this embodiment, the execution sequence of step 13 and step 14 is not necessarily required, and alternatively, step 13 is executed first and then step 14 is executed, or step 14 is executed first and then step 13 is executed, or step 13 and step 14 are executed simultaneously, which is not indispensable, so that the execution sequence of step 13 and step 14 is not limited herein.
Further, in the present embodiment, the second sample entropy (SampEn2) is different from the first sample entropy (SampEn1), and the tolerance γ 2 for calculating the second sample entropy (SampEn2) is also different from the tolerance γ 1 for calculating the first sample entropy (SampEn 1). In particular, the tolerance γ 1 of the calculated first sample entropy (SampEn1) is higher than the tolerance γ 0 of the calculated sample entropy (SampEn0) from the acquired time series of heartbeat intervals of a certain length of time, and the tolerance γ 2 of the calculated second sample entropy (SampEn2) is lower than the tolerance γ 0 of the calculated sample entropy (SampEn0) from the acquired time series of heartbeat intervals of a certain length of time. The following further illustrates the detailed steps of the present embodiment in various embodiments.
FIG. 2 is a flow chart illustrating steps in a method for detecting atrial fibrillation according to an embodiment of the present disclosure. Referring to fig. 2, a method for detecting atrial fibrillation according to an embodiment of the present disclosure includes the following steps.
Step 101: a time series of heartbeat intervals (IBIs) of a certain length of time is taken. Next, step 102 is performed. The reason why the heartbeat interval (IBI) sequence, the R-R wave interval (RRI) sequence or the peak-to-peak interval (PPI) sequence with a certain length of time needs to be obtained is described above and will not be described herein. In one embodiment, the step is implemented by obtaining a time length of the R-R wave interval (RRI) sequence, and the time length of the required R-R wave interval (RRI) sequence is less than 30 seconds, which is equivalent to the required number of samples of the R-R wave interval (RRI) sequence that can be less than 50, so that the method does not require a large amount of data statistics to observe the variability and irregularity of the heartbeat fluctuation, and is more easily applied to a wearable device, and the detection efficiency is improved.
In one embodiment, the step is implemented by using a sequence (or referred to as a difference sequence (△ RR sequence)) formed by differences (△ RR) between adjacent R-R wave intervals (RRI) in a sequence of acquired R-R wave intervals (RRI) of a certain time length, wherein the differences are not accumulated, but dispersed into a difference sequence (△ RR sequence).
Step 103: a first sample entropy tolerance of the detrended time series is calculated. Step 104 is then performed. In this embodiment, the tolerance γ 1 of the calculated first sample entropy (SampEn1) is higher than the tolerance γ 0 of the sample entropy (SampEn0) calculated according to the acquired time series of heartbeat intervals of a certain time length.
Step 104: a first sample entropy of the detrended time series is calculated. Step 105 is then performed. In this embodiment, the calculation of the first sample entropy (SampEn1) is not limited herein as long as the calculated first sample entropy (SampEn1) is obtained according to the tolerance γ 1 calculated in step 103 and can effectively distinguish between atrial fibrillation (AFib) and sinus rhythm (NSR).
Step 105: atrial fibrillation (AFib) and sinus rhythm (NSR) were distinguished by the first sample entropy.
Step 106: the tolerance of a second sample entropy of the detrended time series is calculated. Step 107 is then performed. In this embodiment, the tolerance γ 2 of the calculated second sample entropy (SampEn2) is lower than the tolerance γ 0 of the calculated sample entropy (SampEn0) according to the acquired time series of heartbeat intervals of a certain time length.
Step 107: a second sample entropy of the detrended time series is calculated. Step 108 is then performed. In this embodiment, the calculation of the second sample entropy (SampEn2) is not limited herein as long as the calculated second sample entropy (SampEn2) is obtained according to the tolerance γ 2 calculated in step 106 and can effectively distinguish atrial fibrillation (AFib) from Arrhythmia (ARR) other than sinus rhythm.
Step 108: atrial fibrillation (AFib) and arrhythmia other than sinus rhythm (ARR) are distinguished by a second sample entropy.
In this embodiment, steps 103 to 105 and steps 106 to 108 are performed separately but simultaneously, so that the purpose of distinguishing atrial fibrillation (AFib), sinus rhythm (NSR) and Arrhythmia (ARR) simultaneously can be achieved.
FIG. 3 is a flow chart illustrating steps in a method for detecting atrial fibrillation according to another embodiment of the present application. Referring to fig. 3, another embodiment of the present application provides a method for detecting atrial fibrillation.
Step 201: a time series of heartbeat intervals (IBIs) of a certain length of time is taken. Next, step 202 is performed. The reason why the heartbeat interval (IBI) sequence, the R-R wave interval (RRI) sequence or the peak-to-peak interval (PPI) sequence with a certain length of time needs to be obtained is described above and will not be described herein. In one embodiment, the step is implemented by obtaining a time length of the R-R wave interval (RRI) sequence, and the time length of the required R-R wave interval (RRI) sequence is less than 30 seconds, which is equivalent to the required number of samples of the R-R wave interval (RRI) sequence that can be less than 50, so that the method does not require a large amount of data statistics to observe the variability and irregularity of the heartbeat fluctuation, and is more easily applied to a wearable device, and the detection efficiency is improved.
In one embodiment, the step is implemented by using a sequence (or referred to as a difference sequence (△ RR sequence)) formed by differences (△ RR) between adjacent R-R wave intervals (RRIs) in an obtained sequence of R-R wave intervals (RRIs) of a certain time length to form a difference sequence (△ RR sequence) between the adjacent R-R wave intervals (RRIs), so that the trend changes in the sequence of R-R wave intervals (RRIs) of different time points are not accumulated, but are dispersed into a difference sequence (△ RR sequence).
Step 203: the product of the average value of the time series of heartbeat intervals of a certain time length and a first proportion is used as the tolerance of a first sample entropy in the sample entropy evaluation dimension for calculating the time series of the detrended. Next, step 204 is performed.
In this embodiment, the tolerance γ 1 of the first sample entropy (SampEn1) with the dimension of 1 is evaluated as the sample entropy of a sequence (or difference sequence (△ RR sequence)) formed by taking the product of the average value (meanRR) of the R-R wave interval (RRI) sequence of a certain length of time and a first ratio as the difference value of adjacent R-R wave intervals (RRIs), and the calculated first sample entropy (SampEn1) is used to distinguish atrial fibrillation (AFib) from sinus rhythm (NSR).
In this embodiment, since the sample entropy evaluation dimension of the R-R wave interval (RRI) sequence before the detrending is 2, and the jitter ratio of the arrangement (pattern) of the 2-point RRI sequence and the 3-point RRI sequence in the R-R wave interval (RRI) sequence is relatively equivalent to the jitter ratio of the arrangement of the 1-point difference sequence (△ RR sequence) and the 2-point difference sequence (△ RR sequence) in the difference sequence (△ RR sequence), it is suitable to select the evaluation dimension of the first sample entropy (SampEn1) of the detrended difference sequence (△ RR sequence) as 1.
On the other hand, in the present embodiment, the tolerance γ 1 of the first sample entropy (SampEn1) of the difference sequence of the adjacent RRIs may be selected to be 0.034(0.34X 0.01) to 0.085(0.34X0.25) times of the average value (merr) of the R-R wave pitch (RRI) sequence when the object of calculating the sample entropy shifts to the difference sequence of the adjacent RRIs, based on the tolerance γ 0 of the sample entropy (SampEn) of the general calculation R-R wave pitch (RRI) sequence being 0.01 to 0.25 times of the standard difference of the R-R wave pitch (RRI) sequence and the standard difference of the difference sequence of the adjacent RRIs of atrial fibrillation (AFib) being 0.34 times of the average value (mean RR) of the R-R wave pitch (RRI) sequence. By this step, the tolerance γ 1 for calculating the first sample entropy (SampEn1) may be increased so that the arrhythmia of the sinus rhythm (NSR) may be ignored and not included in the calculation of the first sample entropy (SampEn 1).
Step 204: a first sample entropy of the detrended time series is calculated. Step 205 is then performed. In this embodiment, the calculation of the first sample entropy (SampEn1) is not limited herein as long as the calculated first sample entropy is obtained according to the tolerance γ 1 calculated in step 203 and can effectively distinguish between atrial fibrillation (AFib) and sinus rhythm (NSR).
Step 205: atrial fibrillation (AFib) and sinus rhythm (NSR) were distinguished by the first sample entropy.
FIG. 4A is a graph of a trending NSR R-R wave interval (RRI) sequence using the method of detecting atrial fibrillation according to the embodiments of the present disclosure. FIG. 4B is a difference sequence diagram of R-R wave Spacings (RRIs) in the R-R wave spacing (RRI) sequence of the trending sinus rhythm (NSR) of FIG. 4A. As shown in fig. 4A, the fluctuation of the RRI sequence of the pre-detrended sinus rhythm (NSR) varies from 0.8 seconds to 0.55 seconds, and the trend within this time interval varies so much that the difference between the magnitudes of the RRI sequences at different time points will be incorporated into the sample entropy calculation. Whereas, as shown in fig. 4B, the majority of the fluctuation variations of the difference series of sinus rhythm (NSR) detrended and sample entropy evaluation dimension selected to be 1 were shifted from 0.02 seconds to-0.02 seconds, significantly, the effect of the trending on the sample entropy calculation was greatly reduced. On the other hand, when the tolerance γ 1 for calculating the first sample entropy (SampEn1) of the difference sequence is selected as 0.045 second, which is the product of the average value (mean RR) of the RRI sequence of 0.66 seconds and a ratio of 0.034 to 0.085, e.g., 0.068, it is apparent that most of the fluctuations in fig. 4B are lower than the tolerance γ 1. As such, the arrhythmia of sinus rhythm (NSR) in fig. 4B may be ignored and not accounted for in the calculation of the first sample entropy (SampEn1), such that the calculated first sample entropy (SampEn1) may be used to distinguish atrial fibrillation (AFib) from sinus rhythm (NSR).
Step 206: and taking the product of the standard deviation of the time series subjected to the de-trend and a second proportion as the tolerance for calculating a second sample entropy of the time series subjected to the de-trend in the same sample entropy evaluation dimension. Then, step 207 is performed.
The application aims to solve the problem that the tolerance of the calculated sample entropy (SampEn) of the obtained time sequence of the heartbeat interval (IBI) of the Arrhythmia (ARR) except sinus rhythm (NSR) is underestimated and the low sample entropy characteristic of the Arrhythmia (ARR) except sinus rhythm (NSR) cannot be accurately reflected, so that the Arrhythmia (ARR) except sinus rhythm (NSR) is difficult to be distinguished from atrial fibrillation (AFib), and the sample entropy of the Arrhythmia (ARR) is calculated by the tolerance of the sample entropy (SampEn) of the Arrhythmia (ARR) level, so that the Arrhythmia (ARR) except sinus rhythm (NSR) and the atrial fibrillation (AFib) are distinguished.
In this embodiment, the product of the standard deviation of the sequence (or difference sequence (△ RR sequence)) formed by the difference between adjacent RRIs in the acquired RRI sequence of a certain time length and the second ratio is 0.1 to 0.25 is used as the tolerance γ 2 for calculating the second sample entropy (SampEn2) of the difference sequence (△ RR sequence) of adjacent RRIs, where the evaluation dimension of the second sample entropy (SampEn2) is also 1, and the obtained second sample entropy (SampEn2) is used to distinguish atrial fibrillation (AFib) from Arrhythmia (ARR) other than sinus rhythm.
In this embodiment, the estimated dimension of the second sample entropy (SampEn2) of the detrended difference sequence (△ RR sequence) is still selected to be 1, and may also be 2 or 3, for the same reason as above, on the other hand, for some Arrhythmias (ARR) which may have a higher variation or standard deviation of RRI sequence fluctuations than atrial fibrillation (AFib), but since fluctuations have a specific arrangement (pattern), should be maintained in the estimation category of low irregularity, in other words, if the calculated sample entropy is selected too low, the low irregularity of the Arrhythmia (ARR) cannot be effectively estimated, and the Arrhythmia (ARR) and atrial fibrillation (AFib) cannot be effectively distinguished, in this embodiment, the calculation of the difference sequence of adjacent RRIs (△ RR sequence) is not effectively estimated in the same way as the second sample entropy (SampEn2) of which has the same evaluation dimension 1 of the second sample entropy (SampEn2) and is not reflected by the difference of the first sample entropy (SampEn2) of the first sample entropy (SampEn2) being 0.01, but the estimated by the difference of the second sample entropy (SampEn 34 RR sequence (ARR) of the second sample entropy (sampy) of 0.3. thus, the calculated as the difference of the irreparable sequence (ARR) may not be effectively estimated by 0.0.0.3, 0.3, which is not reflected by the difference of the calculated by the average of the calculated difference of the first sample entropy (ARR sequence of the calculated as the first sample entropy (ARR) of the calculated by the second sample entropy (ARR sequence (ARR) of the average of the second sample entropy (ARR) of the Arrhythmia (ARR) of the calculated by 0.3, which is not 0.3.
Step 207: a second sample entropy of the detrended time series is calculated. Next, step 208 is performed. In this embodiment, the calculation of the second sample entropy (SampEn2) is not limited herein as long as the calculated second sample entropy (SampEn2) is obtained according to the tolerance γ 2 calculated in step 206 and can effectively distinguish atrial fibrillation (AFib) from Arrhythmia (ARR) other than sinus rhythm.
Step 208: atrial fibrillation (AFib) and arrhythmia other than sinus rhythm (ARR) are distinguished by a second sample entropy.
In this embodiment, steps 203-205 and steps 206-208 are performed separately but simultaneously, so that the purpose of distinguishing atrial fibrillation (AFib), sinus rhythm (NSR) and Arrhythmia (ARR) simultaneously can be achieved.
FIG. 5 is a diagram of RRI sequence for other Arrhythmia (ARR) with a certain arrangement of fluctuations and not complete randomness for applying the method of detecting atrial fibrillation according to the embodiments of the present application. As shown in fig. 5, the fluctuating changes in the RRI sequence of the heart arrhythmia before detrending (ARR) are not completely random with an arrangement (pattern). If the tolerance of the sample entropy of the RRI sequence used to calculate the Arrhythmia (ARR) was chosen in the past to be 0.2 times the standard deviation of the RRI sequence, i.e. 0.24X 0.2 times the mean (mean RR) of the RRI sequence 0.6 seconds, i.e. 30 microseconds, then it is significantly lower than the difference between adjacent RRIs. In contrast, in the present embodiment, the product of the standard deviation and a second ratio of the detrended time series is used as a tolerance for calculating a second sample entropy of the detrended time series in the same evaluation dimension of the sample entropy as the first sample entropy, so as to avoid underestimating the tolerance for calculating the sample entropy and to distinguish arrhythmia other than atrial fibrillation and sinus rhythm by the calculated second sample entropy.
In summary, the present application provides a method for detecting atrial fibrillation by converting a time series of heart beat intervals (IBIs) into a de-trended time series before detecting atrial fibrillation using the time series of heart beat intervals (IBIs). The length of time for a sequence of so-called heartbeat intervals (IBIs) may be less than the length of time typically required to detect atrial fibrillation. So-called detrending also includes methods for deriving high frequency fluctuations in the sequence of heart beat intervals by low pass filtering, median filtering, high pass filtering or other similar means. Furthermore, for detrended time series, two different Sample entropies (Sample entropies) are used to distinguish atrial fibrillation (AFib), sinus rhythm (NSR) and other Arrhythmia (ARR), wherein one of the calculated Sample entropies has a higher tolerance than the Sample entropies calculated from the acquired time series of heart intervals of a certain time length, and the other has a lower tolerance than the Sample entropies calculated from the acquired time series of heart intervals of a certain time length, in such a way that in case of a large trend change of the sinus rhythm (NSR), the difference between the magnitudes of the RRI series at different time points is taken into the Sample Entropy (SampEn) calculation and the Arrhythmia (ARR) that may be misjudged as atrial fibrillation (AFib) is excluded. Thus, the proposed method can effectively distinguish between atrial fibrillation (AFib), sinus rhythm (NSR) and other Arrhythmias (ARR) without requiring a sequence of most heart beat intervals.
The detailed description is to be construed as exemplary only and does not limit the scope of the claims, which are appended hereto.

Claims (10)

1. A method of detecting atrial fibrillation, comprising:
obtaining a time sequence of heartbeat intervals of a certain time length;
converting the time series of heartbeat intervals of the certain time length into a de-trended time series;
taking the product of the average value of the time series of heartbeat intervals of the certain time length and a first proportion as the tolerance of calculating a first sample entropy of the detrended time series under a sample entropy evaluation dimension, and distinguishing atrial fibrillation from sinus rhythm by the first sample entropy; and
calculating a tolerance of a second sample entropy of the detrended time series in the sample entropy evaluation dimension as a product of a standard deviation of the detrended time series and a second ratio, and discriminating atrial fibrillation from arrhythmia other than sinus rhythm at the second sample entropy;
wherein the time length of the time series of heartbeat intervals of the certain time length is less than 30 seconds, the tolerance for calculating the second sample entropy is different from the tolerance for calculating the first sample entropy, and the sample entropy evaluation dimension is at most 3.
2. The method of claim 1, wherein the first ratio is 0.034 to 0.085.
3. The method of claim 1, wherein the second ratio is 0.1 to 0.25.
4. The method of claim 3, wherein the second ratio is 0.2.
5. The method of claim 1, wherein the time series of heart beat intervals of a certain length of time is a R-R wave interval series, the detrended time series is a series of differences between adjacent R-R wave intervals of the R-R wave interval series, and the sample entropy evaluation dimension is 1.
6. The method of detecting atrial fibrillation according to claim 1, wherein the step of converting the time series of heart beat intervals of a certain length into the detrended time series further includes filtering out heart beat fluctuations having a frequency of less than 0.4 HZ.
7. A method of detecting atrial fibrillation, comprising the steps of:
obtaining a time sequence of heartbeat intervals of a certain time length;
converting the time series of heartbeat intervals of the certain time length into a de-trended time series;
calculating a first sample entropy of the detrended time series, and distinguishing atrial fibrillation from sinus rhythm with the first sample entropy; and
calculating a second sample entropy of the detrended time series, and discriminating atrial fibrillation from arrhythmia other than sinus rhythm with the second sample entropy;
wherein the tolerance for calculating the second sample entropy is different from the tolerance for calculating the first sample entropy, the tolerance for calculating the first sample entropy is higher than the tolerance for calculating the sample entropy of the time series of heartbeat intervals of the certain length of time, and the tolerance for calculating the second sample entropy is lower than the tolerance for calculating the sample entropy of the time series of heartbeat intervals of the certain length of time.
8. The method of detecting atrial fibrillation according to claim 7, wherein the step of converting the time series of heart beat intervals of a certain length into the detrended time series further includes filtering out heart beat fluctuations having a frequency of less than 0.4 HZ.
9. The method of claim 7, wherein the time series of heart beat intervals is a R-R wave interval series, and the detrended time series is a difference between adjacent R-R wave intervals of the R-R wave interval series.
10. The method of claim 7, wherein the temporal sequence of temporal heartbeat intervals is less than 30 seconds long.
CN201811559121.4A 2018-12-20 2018-12-20 Method for detecting atrial fibrillation Pending CN111345798A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811559121.4A CN111345798A (en) 2018-12-20 2018-12-20 Method for detecting atrial fibrillation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811559121.4A CN111345798A (en) 2018-12-20 2018-12-20 Method for detecting atrial fibrillation

Publications (1)

Publication Number Publication Date
CN111345798A true CN111345798A (en) 2020-06-30

Family

ID=71188165

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811559121.4A Pending CN111345798A (en) 2018-12-20 2018-12-20 Method for detecting atrial fibrillation

Country Status (1)

Country Link
CN (1) CN111345798A (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101150981A (en) * 2005-02-25 2008-03-26 约瑟夫·威塞尔 Method and apparatus for detecting atrial fibrillation
US20170319090A1 (en) * 2009-11-03 2017-11-09 Vivaquant Llc Method and Apparatus for Detection of Heartbeat Characteristics
US20180110432A1 (en) * 2014-12-12 2018-04-26 Soonchunhyang University Industry Academy Cooperation Foundation Method for automatic detection of chf and af with short rr interval time series using electrocardiogram

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101150981A (en) * 2005-02-25 2008-03-26 约瑟夫·威塞尔 Method and apparatus for detecting atrial fibrillation
US20170319090A1 (en) * 2009-11-03 2017-11-09 Vivaquant Llc Method and Apparatus for Detection of Heartbeat Characteristics
US20180110432A1 (en) * 2014-12-12 2018-04-26 Soonchunhyang University Industry Academy Cooperation Foundation Method for automatic detection of chf and af with short rr interval time series using electrocardiogram

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
BINBIN XU等: "Nonlinear trend removal should be carefully performed in heart rate variability analysis", 《ARXIV:1605.05891》 *
LINA ZHAO等: "A New Entropy-Based Atrial Fibrillation Detection Method for Scanning Wearable ECG Recordings", 《ENTROPY》 *
RAÚL ALCARAZA 等: "Optimal parameters study for sample entropy-based atrial fibrillation organization analysis", 《COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE》 *
S. HARGITTAI: "Is it Possible to Detect Atrial Fibrillation by Simply Using RR Intervals?", 《COMPUTING IN CARDIOLOGY 2014》 *

Similar Documents

Publication Publication Date Title
US8233972B2 (en) System for cardiac arrhythmia detection and characterization
US7596405B2 (en) Atrial fibrillation detection
US8396541B2 (en) Signal analysis of cardiac and other patient medical signals
US8755876B2 (en) Methods and systems for atrial fibrillation detection
US5277189A (en) Method and apparatus for the measurement and analysis of cardiac rates and amplitude variations
US8774908B2 (en) Systems for detecting cardiac arrhythmias
US20080269626A1 (en) False positive reduction in spo2 atrial fibrillation detection using average heart rate and nibp
US7092751B2 (en) Detection of atrial arrhythmia
US8868168B2 (en) System for cardiac condition characterization using electrophysiological signal data
US8364248B2 (en) System for cardiac pathology detection and characterization
US8684942B2 (en) System for cardiac impairment detection based on signal regularity
US9050014B2 (en) System for cardiac arrhythmia detection and characterization
US8457724B2 (en) System for heart performance characterization and abnormality detection
US20150126884A1 (en) System for Heart Performance Characterization and Abnormality Detection
EP2895060A1 (en) Screening procedure for identifying risk of arrhythmia
WO2019225662A1 (en) Instantaneous heartbeat reliability evaluation device, method, and program
JP2020517337A (en) Artifact resistance pulse variability measurement
JP2016202345A (en) Biological information processing system, biological information processing device, and analysis result information generation method
CN111345798A (en) Method for detecting atrial fibrillation
Jindal et al. MATLAB based GUI for ECG arrhythmia detection using Pan-Tompkin algorithm
CN115770054A (en) Electrocardiosignal processing method, portable electrocardiosignal acquisition equipment and storage medium
KR20130087940A (en) An hrv detecting method and the hrv detection device
FERRANTI et al. Atrial fibrillation detection in PPG signal recorded through a wristband device
KR101515725B1 (en) Apparatus and method for diagnosing cardiac disorder
Zheng et al. T-wave alternans: A comparison of different measurement techniques

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
AD01 Patent right deemed abandoned

Effective date of abandoning: 20230224

AD01 Patent right deemed abandoned