Elsevier

Physiology & Behavior

Volume 101, Issue 1, 4 August 2010, Pages 74-80
Physiology & Behavior

Applying fractal analysis to heart rate time series of sheep experiencing pain

https://doi.org/10.1016/j.physbeh.2010.04.018Get rights and content

Abstract

The objective assessment of pain is difficult in animals and humans alike. Detrended fluctuation analysis (DFA) is a method which extracts “hidden” information from heart rate time series, and may offer a novel way of assessing the subjective experience associated with pain. The aim of this study was to investigate whether any fractal differences could be detected in heart rate time series of sheep due to the infliction of ischaemic pain. Heart rate variability (HRV) was recorded continuously in five ewes during treatment sequences of baseline, intervention and post-intervention for up to 60 min. Heart rate time series were subjected to a DFA, and the median of the scaling coefficients (α) was found to be α = 1.10 for the baseline sequences, 1.01 for the intervention sequences and 1.00 for the post-intervention sequences. The complexity in the regulation of heartbeats decreased between baseline and intervention (p  0.03) and baseline and post-intervention (p  0.01), indicating reperfusion pain and nociceptive sensitization in the post-intervention sequence. Random time series based on Gaussian white noise were generated, with similar mean and variance to the HRV sequences. No difference was found between these series (p  0.28), pointing to a true difference in complexity in the original data. We found no difference in the scaling coefficient α between the different treatments, possibly due to the small sample size or a fear induced sympathetic arousal during test day 1 confounding the results. The decrease in the scaling coefficient α may be due to sympathetic activation and vagal withdrawal. DFA of heart rate time series may be a useful method to evaluate the progressive shift of cardiac regulation toward sympathetic activation and vagal withdrawal produced by pain or negative emotional responses such as fear.

Introduction

Quantifying the degree of pain experienced by animals is an important component when assessing animal welfare [1]. The objective assessment of pain in sheep is difficult, particularly because overt behavioural responses in this species are limited. Non-invasive, dynamic, real-time measures of sympathovagal activity may offer a novel way of assessing the subjective experience associated with pain. The sheep heart is similar to that of a human in many ways, including dimensions of the chambers, coronary anatomy, and magnitude of hemodynamic variables such as blood pressure, heart rate and cardiac output. Autonomic innervations of the sheep heart are also similar to that of a human [2]. Comparative studies have therefore been performed in sheep [3].

Physiologic systems generate complex fluctuations in their output signals that reflect the underlying dynamics. Three particularly vexing features of physiological time series such as heart rates are non-stationarity, nonlinearity and nonequilibrium phenomena [4]. Heart rate variability (HRV) is measured by determining the constantly changing temporal distance between consecutive heartbeats (RR intervals), and is an integrative measurement variable that reflects the prevailing balance of vagal and sympathetic tone [5].

It has been suggested that the time series from heart rate recordings contain hidden information, which is not extractable with conventional methods of analysis [6]. Detrended fluctuation analysis (DFA) is a recent approach for extracting such information from physiological time series. The DFA method has been used to detect long-range or long-term correlation in non-stationary time series in various physiological and pathological conditions, including the activation of the autonomic nervous system [7], ventricular fibrillation [8] and dilated cardio-myopathy [9].

The highly irregular behaviour of cardiac interbeat intervals defies conventional analyses that require “well-behaved,” stationary data sets [6]. Time and frequency domain HRV measures attempt to quantify HRV on various time scales. These traditional measures can be complemented by nonlinear HRV measures, which attempt to quantify the structure or complexity of the RR interval time series [10].

Normal heart rate time series have been shown to be fractal-like and self-similar, defined as when a subunit of RR interval time series resembles the larger time scale. This indicates long-range correlation between RR intervals, which means that interbeat intervals are partially dependent on the intervals at previous points [11]. Using the DFA method, such correlated, fractal-like behaviour can be measured with the use of a ‘scaling coefficient’ known as α[10]. A scaling coefficient α value of 0.5 (white noise) implies no correlation between the RR intervals as a result of random heart rate dynamics. A scaling coefficient value of 1.0 (1/f or pink noise) contains both random and highly correlated characteristics in RR interval time series and indicates fractal heart rate dynamics [11]. The Brownian noise has a value of 1.5 and is the integration of the white noise [10]. A change from pink noise towards more random fluctuations (white noise) with no correlation between interbeat intervals has been shown to be physiologically deleterious, and this breakdown in long-range correlations occur during atrial fibrillation over relatively short time scales [6], [12]. The DFA algorithm quantifies fractal-like correlation properties by calculating the scaling property of the root-mean-square fluctuation of the integrated and detrended time series data [4].

Pain induces systemic responses that affect physiological processes such as heart rate. A decrease in HRV parameters has been found to be associated with the progress of laminitis, a disease that leads to considerable pain in horses [13]. Time domain HRV parameters were also found to be decreased during mild to moderate pain in sheep [14].

The tourniquet pain model has previously been used in sheep, and the administration of the non-steroidal anti-inflammatory drug (NSAID) flunixin meglumine (1 mg/kg) has been found to reduce hyperalgesia in sheep during a noxious ischemic stimulus [15]. Theories on why prolonged tourniquet inflation produces pain have supported the role of C-fibers in the transmission of tourniquet-related pain [16]. It is hypothesized that the pain resulting from tourniquet-related nerve compression and ischaemic pain may involve several changes in peripheral afferent input and in the threshold of spinal dorsal horn neurons [17].

Traditional time series analysis gives little or no information regarding the fractality of the sequence of points studied. Hence, it is difficult to detect underlying structural differences in HRV data using standard time series analysis [12]. Our aim was therefore to investigate whether the DFA method could detect fractal differences in the heart rate time series of sheep when subjected to sustained, mild to moderate pain infliction.

Section snippets

Materials and methods

This study was conducted as part of a larger study which aimed to explore HRV and infrared thermography as non-invasive methods to assess pain in sheep [14].

Results

No association was found between the different treatments and the scaling coefficient α (ANOVA, p  0.5). Using linear regression, we found differences of the scaling coefficient α between baseline and intervention (p < 0.03) and baseline and post-intervention (p < 0.01). No difference was found between the intervention and post-intervention sequences (p  0.75). By adding each ewe as a factor, the variance explained of the regression model increased from adjusted R2 = 0.10, p  0.35 to adjusted R2 = 0.19, p 

Discussion

We reported previously that linear parameters of HRV (RMSSD; the root mean square of successive RR intervals and SDNN; the standard deviation of all interbeat intervals) provided a method sufficiently sensitive to detect the presence of mild to moderate ischaemic pain in sheep [14]. The time domain parameters were found to be decreased during the noxious ischaemic stimulus (S) treatment. Complementary to our previous work [14], in the present study we report that the DFA method was able to

Acknowledgements

The authors wish to thank Eystein Glattre for interesting discussions. The Norwegian Research Council, NORTURA and ‘Fondet for forskningsavgift på landbruksprodukter’ are acknowledged for funding this study.

References (43)

  • P. Jensen et al.

    Stress as a state of motivational systems

    Appl Anim Behav Sci

    (1997)
  • K. Carlsson et al.

    Predictability modulates the affective and sensory-discriminative neural processing of pain

    NeuroImage

    (2006)
  • J.L. Rhudy et al.

    Fear and anxiety: divergent effects on human pain thresholds

    Pain

    (2000)
  • A. Romeyer et al.

    Assessment of fear reactions in domestic sheep, and influence of breed and rearing conditions

    Appl Anim Behav Sci

    (1992)
  • L. Greiveldinger et al.

    Emotional experience in sheep: predictability of a sudden event lowers subsequent emotional responses

    Physiol Behav

    (2007)
  • J.L. Barnett

    Measuring pain in animals

    Aust Vet J

    (1997)
  • R.G. Yeh et al.

    Detrended fluctuation analyses of short-term heart rate variability in surgical intensive care units

    Biomed Eng — Appl Basis Commun

    (2006)
  • E. Mohr et al.

    Heart rate variability—a noninvasive approach to measure stress in calves and cows

    Physiol Behav

    (2006)
  • C.-K. Peng et al.

    Nonlinear dynamics, self-organization, and biomedicine

  • T.H. Mäkikalli et al.

    Heart rate dynamics before spontaneous onset of ventricular fibrillation in patients with healed myocardial infarcts

    Am J Cardiol

    (1999)
  • N.G. Mahon et al.

    Fractal correlation properties of RR interval dynamics in asymptomatic relatives of patients with dilated cardiomyopathy

    Eur J Heart Fail

    (2002)
  • Cited by (13)

    • Fractal measures in activity patterns: Do gastrointestinal parasites affect the complexity of sheep behaviour?

      2018, Applied Animal Behaviour Science
      Citation Excerpt :

      Indeed, reduced variability in space or over time through ageing and disease has been demonstrated in many human physiological, cognitive and motor systems (Lipsitz and Goldberger, 1992; Ribeiro et al., 2002; Vaillancourt and Newell, 2002; Benoit et al., 2011). Among domestic animals, lower complexity (i.e. greater periodicity or stereotypy) in temporal dynamics has been reported in the heart rate of sheep experiencing pain (Stubsjøen et al., 2010) and of swine during hemorrhagic shock (Batchinsky et al., 2007). Concerning animal behaviour, fractal analyses, in particular, have revealed lowered complexity in behaviours of stressed or physiologically challenged individuals, in numerous species (see reviews in Seuront, 2010, 2015; MacIntosh, 2014; Seuront and Cribb, 2017).

    • Assessment of chronic stress in sheep (part II): Exploring heart rate variability as a non-invasive measure to evaluate cardiac regulation

      2015, Small Ruminant Research
      Citation Excerpt :

      RMSSD: sqrt(yij) = a0 + bjxj + eijSDNN: log(yij) = a0 + bjxj + eijLF/HF: log(yij) = a0 + bjxj + eijDFA: yij = a0 + bjxj + eij The DFA based α measure was calculated as described in Stubsjøen et al. (2010). Two α values were removed from the study due to unrealistic low values (α < 0.6).

    • Comparison of wave height interpolation with wavelet refined cubic spline and fractal methods

      2014, Ocean Engineering
      Citation Excerpt :

      The concept of fractal was first introduced by Mandelbrot (1967) in order to illustrate irregular patterns that cannot be analysed by traditional Euclidean geometry, and one of the most commonly used examples to illustrate fractal is the coastline (Jiang and Plotnick, 1998; Mandelbrot, 1967; Phillips, 1986; Schwimmer, 2008). The fractal method has also been applied in the analysis of the correlation in the time series for various studies, from physiology (Stubsjøen et al., 2010), economy (Muzy et al., 2000) to hydrology (Jayawardena and Lai, 1994). For wave research, Reikard (2009) forecasted wave energy by determining the wave parameter׳s fractal dimension, but the concept of fractal was not directly involved in the prediction.

    View all citing articles on Scopus
    View full text