+ All Categories
Home > Documents > Instantaneous Monitoring of Heart Rate Variability

Instantaneous Monitoring of Heart Rate Variability

Date post: 11-Jan-2016
Category:
Upload: kale
View: 38 times
Download: 6 times
Share this document with a friend
Description:
Instantaneous Monitoring of Heart Rate Variability. Outline. Abstract Introduction Methodology Results. Abstract. - PowerPoint PPT Presentation
Popular Tags:
22
Transcript
Page 1: Instantaneous Monitoring of Heart Rate  Variability
Page 2: Instantaneous Monitoring of Heart Rate  Variability

OutlineAbstractIntroductionMethodologyResults

Page 3: Instantaneous Monitoring of Heart Rate  Variability

AbstractMost of the currently accepted approaches to compute heart rate and assess heart rate variability operate on interpolatedinterpolated, continuous-valued heart rate signalscontinuous-valued heart rate signals, thereby ignoring the underlying discrete structure of human heart beats.

To overcome this limitation, we model the stochastic structure of heart beat intervals as a history-dependent, inverse Gaussian process and derive from it an explicit probability density describing heart rate and heart rate variability.

Page 4: Instantaneous Monitoring of Heart Rate  Variability

We estimate the parameters of the inverse Gaussian model by local maximum likelihood and assess model goodness-of-fit using Q-Q plot analyses (Quantile-Quantile Normal Plots-常態分位數圖 )goodness-of-fit ( 適合度 ):此種檢定是看我們之實際值 ( 或觀測值 ) 是否服從某一理論之分配。這種實際值(或觀測值)與理論值之間之配合程度之檢定問題稱之為適合度檢定。

Page 5: Instantaneous Monitoring of Heart Rate  Variability

We apply our model in an analysis of human heart beat intervals from a tilt-table experiment.

Page 6: Instantaneous Monitoring of Heart Rate  Variability

IntroductionIn the last 40 years, heart rate (HR) and heart rate variability (HRV) have been established as important quantitative indices of cardiovascular control by the autonomic nervous system, as well as effective diagnostic tools and predictors of mortality for diseases related to cardiovascular function and regulation

Page 7: Instantaneous Monitoring of Heart Rate  Variability

HR is the number of R-wave events (heart beats) per unit time on the electrocardiogram (ECG).

HRV is defined as the variation in the R-R intervals, i.e., in the times between the R-wave events.

Neither HR nor HRV can be observed directly from the ECG, but both must be estimated from the sequence of R-R intervals

Page 8: Instantaneous Monitoring of Heart Rate  Variability

There are several methodological limitations to current methods used to estimate HRV.

In research studies, current time domain, frequency domain, dynamical systems, and entropy methods for HRV analysis generally require several minutes or more of ECG measurements in order to produce meaningful analyses, and these data often must be collected under stationary conditions.

Page 9: Instantaneous Monitoring of Heart Rate  Variability

In addition, most of these methods must convert R-R interval data into evenly spaced, continuous-valued measurements for analysis by first interpolating the HR series estimate computed from either the local averages model or the reciprocal model

While all of these methods give important characterizations of human heart beat dynamics, none provides a goodness-of-fit assessment to measure how well the R-R interval data are described by a particular model.

Page 10: Instantaneous Monitoring of Heart Rate  Variability

In response to these shortcomings, we present a new statistical framework that models the R-wave events as a discrete event defined by an inverse Gaussian parametric probability function

As such, our approach is able to avoid the need for conversion to continuous-valued signals and is also able to formally assess model goodness-of-fit through well established techniques for comparing discrete events models.

Page 11: Instantaneous Monitoring of Heart Rate  Variability

MethdologyA. Heart Rate Probability Model

• In an observation interval (0,T]• 0 <U1 <U2 <...,<U5 ..., <Un ≦ T as the N

successive R-wave event times detected from an ECG.

• Hun is the history of the R-R intervals up to Un

• θ is a set of p model parameters

Page 12: Instantaneous Monitoring of Heart Rate  Variability

The history term represents the influence of recent parasympathetic and sympathetic inputs to the SA node on the R-R interval length by modeling the mean as a linear function of the previous R-R intervals.

The Mean and Standard deviation of the R-R interval probability model in (1) are , respectively:

Page 13: Instantaneous Monitoring of Heart Rate  Variability

高斯分佈

早在 18世紀就有數學家和天文學家開始探討這樣的一條曲線。德國天文家兼數學家高斯( Carl Friedrich Gauss, 1777-1855)利用常態分佈研究天文學觀察中誤差的分佈情形,因此常態分佈又稱高斯分佈。

另一位著名的數學和統計學家 Karl Pearson( 1857-1936)將高斯分佈稱為常態分佈。

Page 14: Instantaneous Monitoring of Heart Rate  Variability

0

10

20

30

40

50

60

70

80

90

150 155 160 165 170 175 180 185 190

身 高

人 數

Page 15: Instantaneous Monitoring of Heart Rate  Variability

這條曲線的數學函數為

其中 p = 3.1416 , e 是自然對數之底2.7183 , X 介在正負無限大, m 是平均數, s 是標準差。一旦確定平均數和標準差後,帶入公式算得f(X) 。

2

2

12

2

1,;

X

eXfY

Page 16: Instantaneous Monitoring of Heart Rate  Variability

要決定常態分佈的形狀,就必須知道平均數 m 和變異數 s2

(或者標準差 s )。常態分佈取決於兩個參數( parameter): m 和 s2 。

只要設定這兩個參數,就可以畫出那條常態分佈曲線。只要 m 或 s2 不同,曲線就不同。

這也就是為何在上述公式裡,表明 其中分號後面代表的就是決定這個函數的參數。假如變數 X 服從常態分佈,平均數為 m ,變異數為 s2 ,則寫成: X ~ N(m, s2) ,其中 ~ 表示服從, N 表示常態分佈。

Page 17: Instantaneous Monitoring of Heart Rate  Variability
Page 18: Instantaneous Monitoring of Heart Rate  Variability
Page 19: Instantaneous Monitoring of Heart Rate  Variability

B. Model Goodness-of-FitBecause the R-R interval model in (1) defines an explicit discrete event model, we can use a quantile-quantile (Q-Q) analysis to evaluate model goodness-of-fit

Page 20: Instantaneous Monitoring of Heart Rate  Variability
Page 21: Instantaneous Monitoring of Heart Rate  Variability

Result

Page 22: Instantaneous Monitoring of Heart Rate  Variability

Fig. 2. Autoregressive spectral estimation of the supine (top panel) and thetilt (bottom panel) segments of the interpolated reciprocal R-R intervals inFig. 1A (dotted line) and our HR estimates in Fig. 1C (solid line).


Recommended