f20080820101055.ppt

47
Finding Hidden Information in heart rate dynamics Men-Tzung Lo, Ph.D, Assistant Research Scientist, Research-Center-Adaptive-Data- Analysis, NCU

Upload: simon23

Post on 13-May-2015

333 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: f20080820101055.ppt

Finding Hidden Information in heart rate dynamics

Men-Tzung Lo, Ph.D,

Assistant Research Scientist,

Research-Center-Adaptive-Data-Analysis, NCU

Page 2: f20080820101055.ppt

PhysioNet

Page 3: f20080820101055.ppt

Between Genomics and Diagnostics Something is Missing…

Biomedical Informatics: Methods, Techniques and Theories

BioinformaticsImaging

InformaticsClinical

InformaticsPublic HealthInformatics

Molecular and Cellular Processes

Tissues and Organs

Populations And Society

Individuals(Patients)

?

Page 4: f20080820101055.ppt

A More Complete Picture

Biomedical Informatics: Methods, Techniques and Theories

BioinformaticsImaging

InformaticsClinical

InformaticsPublic HealthInformatics

Molecular and Cellular Processes

Tissues and Organs

Diagnostic and Functional Dynamics

Populations And Society

Individuals(Patients)

Complex SignalsInformatics

Page 5: f20080820101055.ppt

Which is the Healthy Subject?Escape statistical distinction based on conventional comparisons

Page 6: f20080820101055.ppt

Variability vs. Complexity Ary L Golberberg, “complex system”,

ProC Am Thorac soc

Page 7: f20080820101055.ppt

“Beyond ANOVA” (ANalysis Of VAriance between groups) Three Key Concepts( The purpose of complex signal informatics ):

1. Physiologic signals are the most complex in nature

2. Important basic/clinical information is “hidden” (encoded) in these fluctuations

3. Complexity degrades with pathology/aging

The often “noisy” variability actually is the signal and

represents the nonlinear signaling mechanisms

Page 8: f20080820101055.ppt

Body as servo-mechanism type machine

• Importance of corrective mechanisms to keep variables “in bounds.”

• Healthy system are self-regulated to reduce variability and maintain physiologic constancy.

Underlying notion of “constant,” “steady-state,””

conditions.

Restored steady state

…ORBaseline

Perturbation

Page 9: f20080820101055.ppt

Homeostasis Revisited

…OR

Is complex spatio-temporal variability a mechanism of object with multi-organization ?

But, What’s the healthy complexity ?

Page 10: f20080820101055.ppt

Some Characteristics of Healthy Complexity

• Nonstationarity• Statistics change with time

• Nonlinearity• Components interact in unexpected ways ( “cross-talk”, the

superposition paradigm fails )• Multiscale Organization

• Fluctuations/structures may have fractal organization• Time Irreversibility

• Non-Periodic signal

Healthy Heart Rate Dynamics

Page 11: f20080820101055.ppt

Is Your World Linear or Nonlinear?• Linear Process:

• Simple rules simple behaviors • Things add up• Proportionality of input/output• High predictability, no surprises

• Nonlinear Process:• Simple rules complex behaviors • Small changes may have huge effects• Low predictability & anomalous behaviors• Whole sum of parts

Page 12: f20080820101055.ppt

*** Danger ***

Linear Fallacy: Widely-held assumption that biologicalsystems can be largely understood by dissecting out micro-components or modules and analyzing them in isolation.

“Rube Goldberg physiology”Pencil Sharpener

Page 13: f20080820101055.ppt

“Nonlinear” Pharmacology

Treatment of Chronic Heart Failure Linear (target) approach: increase contractility*• Milrinone • Vesnarinone

Systems approach: interrupt vicious neurohormonal cycle**

• Beta-blockers

* Excess mortality** Enhanced survival

Page 14: f20080820101055.ppt
Page 15: f20080820101055.ppt

• Hypotheses :• The output of physiologic systems often

becomes more regular and predictable with disease

Loss of Complexity/Information with Disease

Page 16: f20080820101055.ppt

Multiscale Time Irreversibility (MTI):

• Time irreversibility is greatest for healthy physiologic dynamics, which have the highest adaptability

• Time irreversibility decreases with aging and disease

Healthiest vs Sickest

Page 17: f20080820101055.ppt

Congestive heart failure

Page 18: f20080820101055.ppt

Heart Rate Fluctuates Cyclically During Sleep Apnea

60 minutes of data

Page 19: f20080820101055.ppt

Complexity analysis is developed to quantize the dynamics of

biology signals

Page 20: f20080820101055.ppt

Wonderful World of “Hidden” Complexity/Nonlinear Mechanisms in Physiology

• Bifurcations (abrupt change) • Nonlinear oscillations• Time asymmetry• Deterministic chaos• Fractals

• Nonlinear waves: spirals/scrolls

• Hysteresis

Biomedical signals that have been analyzed using complex signal informatics include heart rate, nerve activity, renal flow, arterial pressure, and respiratory waveforms.

Page 21: f20080820101055.ppt

Nonlinear Mechanisms in Physiology

• Bad news: physiology is complex!• Good news: the complex behavior can arise in general

mechanisms with simple rules

Page 22: f20080820101055.ppt

Fractal: Complex tree-like object or hierarchical process, composed of sub-units (and sub-sub-units, etc) that resemble the larger scale design.

This internal look-alike property is known asself-similarity or scale-invariance.

Fractals as a Design Principle in Nature

Page 23: f20080820101055.ppt

Fractal Self-Organization:Coronary Artery Tree

Page 24: f20080820101055.ppt

Fractals and Information Transmission:Purkinje Cells in Cerebellum

Page 25: f20080820101055.ppt

Fractal: A tree-like object or process, composed ofsub-units (and sub-sub-units, etc) that resemble thelarger scale structure

Self-similarity (scale invariance), therefore, may be a property of dynamics as well as structure

Fractal dynamics has memory effect (long range correlation: adjust to fit any scales)

Are there Fractal (Scale-Free) Processes in Biology?

Page 26: f20080820101055.ppt

Why is it Physiologic to be Fractal?

• Healthy function requires capability( non-integer fractal dimension) to cope

with unpredictable environments

• Scale-free (fractal) systems generate broad range of long-range correlated responses “memory effect”

disorder

Fractal mechanism

Page 27: f20080820101055.ppt

Fractal Complexity Degrades with Disease

Single Scale Periodicity Uncorrelated Randomness

Two Patterns ofPathologic Breakdown

Healthy Dynamics: Multiscale Fractal Variability

Nature 1999; 399:461Phys Rev Lett 2002; 89 : 068102

Healthy dynamicspoised between too much order and total randomness.

But randomness isnot chaos!

Page 28: f20080820101055.ppt

Transformation

Seems irregular ?

Page 29: f20080820101055.ppt
Page 30: f20080820101055.ppt

ApEn

Page 31: f20080820101055.ppt

Biological systems need to operate across multiple spatial and temporal scales; and hence their complexity is also multiscaled

Page 32: f20080820101055.ppt

DFA & multi-fractal

Page 33: f20080820101055.ppt

Scale dependent fractal (Detrend-Fluctuation-Analysis method, C.-k.Peng,1995, chaos )

• The average root-mean-square fluctuation functions F(n) is obtained after integrating and detrending the data (to exclude environmental stimuli)

Slope =0.5 un-correlation

Anti-correlation

Fractal

Harmonic or periodic

Page 34: f20080820101055.ppt

Color-coded wavelet analysis (Plamen Ch Ivanov, nature ,1999 )

singularity

Page 35: f20080820101055.ppt

Diagnosis To prognosis

Normal

abnormal

Traditional signal Complexity analysis can

help specify the abnormal. But, what is feature for the critical

case?

Page 36: f20080820101055.ppt

Dynamics of heart rate (HR)

Sympathetic stimulationSympathetic stimulation

Parasympathetic stimulationParasympathetic stimulation

HR(bpm)

Heart rate

Heart rate

Page 37: f20080820101055.ppt

Quantification of HR dynamics

• Time domain measurement

• Standard deviation of normal-to-normal beat intervals (SDNN)

• Power spectrum analysis• High frequency (HF)

• Low frequency (LF)

• Very low frequency (VLF)

• LF/HF

Task Force of the European Society of Cardiology and the North American of Pacing and Electrophysiology ,

Circulation 93:1043-65,1996

Page 38: f20080820101055.ppt

• Fourier analysis is a valid technique for investigation of the oscillatory components of circulatory and respiratory systems.

(Attinger, et al., Biophys. J. 6:291-304, 1966.)

Applicability of fourier transform In analysis of biological systems

Transformation

Page 39: f20080820101055.ppt

Quantification of HR dynamics

• Time domain measurement

• Standard deviation of normal-to-normal beat intervals (SDNN)

• Power spectrum analysis• High frequency (HF)

• Low frequency (LF)

• Very low frequency (VLF)

Task Force of the European Society of Cardiology and the North American of Pacing and Electrophysiology ,

Circulation 93:1043-65,1996

Page 40: f20080820101055.ppt

Clinical applications

Task Force of the European Society of Cardiology and the North American of Pacing and Electrophysiology,

Circulation 93:1043-65,1996

Page 41: f20080820101055.ppt

Clinical applications (cont’)

Task Force of the European Society of Cardiology and the North American of Pacing and Electrophysiology ,

Circulation 93:1043-65,1996

Page 42: f20080820101055.ppt

Modulation of heartbeats

Sympathetic nerve Parasympathetic nerve

baroreceptor chemoreceptor

Stretch receptor

Page 43: f20080820101055.ppt

Analysis of HR dynamics by nonlinear methods

• Dynamic measures of HRV may uncover abnormalities that are not easily detectable with traditional time and frequency domain measures.

Laitio, et al., Anesthesiology 93:69-80,2000

Page 44: f20080820101055.ppt

Huikuri, et al. Circulation 101:47–53, 2000

Applications of nonlinear analysis in patients with myocardial infarction

Page 45: f20080820101055.ppt

Application of nonlinear analysis in heart failure patients

Makikallio, et al. Am J Cardiol 87:178–82, 2001

Page 46: f20080820101055.ppt

Quatification of Poincaré plot

Huikuri, et al. Circulation 93:1836-44, 1996

SD2:Long-term HRV

SD1:Instantaneous HRV

SD1/SD2:Shape of the plot

Page 47: f20080820101055.ppt

Thank you for your attention