Quantification of fetal heart rate regularity using symbolic dynamics

A Comprehensive Review of Techniques for Processing and Analyzing Fetal Heart Rate Signals

The availability of standardized guidelines regarding the use of electronic fetal monitoring (EFM) in clinical practice has not effectively helped to solve the main drawbacks of fetal heart rate (FHR) surveillance methodology, which still presents inter- and intra-observer variability as well as uncertainty in the classification of unreassuring or risky FHR recordings. Given the clinical relevance of the interpretation of FHR traces as well as the role of FHR as a marker of fetal wellbeing autonomous nervous system development, many different approaches for computerized processing and analysis of FHR patterns have been proposed in the literature. The objective of this review is to describe the techniques, methodologies, and algorithms proposed in this field so far, reporting their main achievements and discussing the value they brought to the scientific and clinical community. The review explores the following two main approaches to the processing and analysis of FHR signals: traditional (or linear) methodologies, namely, time and frequency domain analysis, and less conventional (or nonlinear) techniques. In this scenario, the emerging role and the opportunities offered by Artificial Intelligence tools, representing the future direction of EFM, are also discussed with a specific focus on the use of Artificial Neural Networks, whose application to the analysis of accelerations in FHR signals is also examined in a case study conducted by the authors.

Comparison of fetal heart rate variability by symbolic dynamics at the third trimester of pregnancy and low-risk parturition

Fetal heart rate variability (fHRV) is an essential source of information to monitor fetal well-being during pregnancy. This study aimed to apply a nonlinear approach, known as symbolic dynamics (SD), for comparing human fHRV in the third trimester of pregnancy during active fetal state (TT) and active labor at term (P). We performed a longitudinal, prospective, descriptive, and comparative study composed of 42 longitudinal recordings of 5-minutes of fetal heartbeat interval series. Recordings were collected from 21 low-risk, healthy, pregnant women attending the Maternal and Child Research Center (CIMIGen), Mexico City. We calculated relevant linear parameters of fHRV between TT and P stages, such as the percentage of differences between adjacent RR intervals >5 ms (PRR5, related to vagal modulations) and other SD parameters such as the percentage of no variations between three successive symbols (%0V, reflects sympathetic modulations) and the probability of low variability with a threshold of 4 ms (POLVAR4, associated with a low variability). We identified statistical differences for PRR5 between TT and P (37.13% [28.47-47.60%] vs. 28.84% [19.36-36.76%], p = 0.03), respectively. Also, for 0V% (65.66% [59.01-71.80%] vs. 71.14% [65.94-75.87%], p = 0.03) and for POLVAR4 values (0.06 [0.04-0.11] vs. 0.15 [0.09-0.24], p = 0.002), respectively. Our results indicate that during parturition, the short-term fetal fHRV is decreased, showing a decreased vagal modulations and higher adrenergic response of the heart. These autonomic modifications may result from the fetal response to the stressful inflammatory challenge of labor. We thus confirmed that the analysis of the SD applied to fHRV time series could be a potential clinical biomarker to differentiate the fetal autonomic cardiac condition at different stages of pregnancy.

Monitoring fetal maturation-objectives, techniques and indices of autonomic function

Monitoring the fetal behavior does not only have implications for acute care but also for identifying developmental disturbances that burden the entire later life. The concept, of 'fetal programming', also known as 'developmental origins of adult disease hypothesis', e.g. applies for cardiovascular, metabolic, hyperkinetic, cognitive disorders. Since the autonomic nervous system is involved in all of those systems, cardiac autonomic control may provide relevant functional diagnostic and prognostic information. The fetal heart rate patterns (HRP) are one of the few functional signals in the prenatal period that relate to autonomic control and, therefore, is predestinated for its evaluation. The development of sensitive markers of fetal maturation and its disturbances requires the consideration of physiological fundamentals, recording technology and HRP parameters of autonomic control. Based on the ESGCO2016 special session on monitoring the fetal maturation we herein report the most recent results on: (i) functional fetal autonomic brain age score (fABAS), Recurrence Quantitative Analysis and Binary Symbolic Dynamics of complex HRP resolve specific maturation periods, (ii) magnetocardiography (MCG) based fABAS was validated for cardiotocography (CTG), (iii) 30 min recordings are sufficient for obtaining episodes of high variability, important for intrauterine growth restriction (IUGR) detection in handheld Doppler, (iv) novel parameters from PRSA to identify Intra IUGR fetuses, (v) evaluation of fetal electrocardiographic (ECG) recordings, (vi) correlation between maternal and fetal HRV is disturbed in pre-eclampsia. The reported novel developments significantly extend the possibilities for the established CTG methodology. Novel HRP indices improve the accuracy of assessment due to their more appropriate consideration of complex autonomic processes across the recording technologies (CTG, handheld Doppler, MCG, ECG). The ultimate objective is their dissemination into routine practice and studies of fetal developmental disturbances with implications for programming of adult diseases.

Symbolic features and classification via support vector machine for predicting death in patients with Chagas disease

This paper introduces a technique for predicting death in patients with Chagas disease using features extracted from symbolic series and time-frequency indices of heart rate variability (HRV). The study included 150 patients: 15 patients who died and 135 who did not. The HRV series were obtained from 24-h Holter monitoring. Sequences of symbols from 5-min epochs from series of RR intervals were generated using symbolic dynamics and ordinal pattern statistics. Fourteen features were extracted from symbolic series and four derived from clinical aspects of patients. For classification, the 18 features from each epoch were used as inputs in a support vector machine (SVM) with a radial basis function (RBF) kernel. The results showed that it is possible to distinguish between the two classes, patients with Chagas disease who did or did not die, with a 95% accuracy rate. Therefore, we suggest that the use of new features based on symbolic series, coupled with classic time-frequency and clinical indices, proves to be a good predictor of death in patients with Chagas disease.

Software for computerised analysis of cardiotocographic traces

Despite the widespread use of cardiotocography in foetal monitoring, the evaluation of foetal status suffers from a considerable inter and intra-observer variability. In order to overcome the main limitations of visual cardiotocographic assessment, computerised methods to analyse cardiotocographic recordings have been recently developed. In this study, a new software for automated analysis of foetal heart rate is presented. It allows an automatic procedure for measuring the most relevant parameters derivable from cardiotocographic traces. Simulated and real cardiotocographic traces were analysed to test software reliability. In artificial traces, we simulated a set number of events (accelerations, decelerations and contractions) to be recognised. In the case of real signals, instead, results of the computerised analysis were compared with the visual assessment performed by 18 expert clinicians and three performance indexes were computed to gain information about performances of the proposed software. The software showed preliminary performance we judged satisfactory in that the results matched completely the requirements, as proved by tests on artificial signals in which all simulated events were detected from the software. Performance indexes computed in comparison with obstetricians' evaluations are, on the contrary, not so satisfactory; in fact they led to obtain the following values of the statistical parameters: sensitivity equal to 93%, positive predictive value equal to 82% and accuracy equal to 77%. Very probably this arises from the high variability of trace annotation carried out by clinicians.

Strategies of symbolization in cardiovascular time series to test individual gestational development in the fetus

The analysis of symbolic dynamics applied to physiological time series retrieves dynamical properties of the underlying regulation which are robust against the symbolic transformation. In this study, three different transformations to produce a symbolic series were applied to fetal RR interval series to test whether they reflect individual changes of fetal heart rate variability in the course of pregnancy. Each transformation was applied to 215 heartbeat datasets obtained from 11 fetuses during the second and the third trimester of pregnancy (at least 10 datasets per fetus, median 17). In the symbolic series, the occurrence of symbolic sequences of length 3 was categorized according to the amount of variations in the sequence: no variation of the symbols, one variation, two variations. Linear regression with respect to gestational age showed that the individual course during pregnancy performed best using a binary transformation reflecting whether the RR interval differences are below or above a threshold. The median goodness of fit of the individual regression lines was 0.73 and also the variability among the individual slopes was low. Other transformations to symbolic dynamics performed worse but were still able to reflect the individual progress of fetal cardiovascular regulation.

Development of multiscale complexity and multifractality of fetal heart rate variability

During fetal development a complex system grows and coordination over multiple time scales is formed towards an integrated behavior of the organism. Since essential cardiovascular and associated coordination is mediated by the autonomic nervous system (ANS) and the ANS activity is reflected in recordable heart rate patterns, multiscale heart rate analysis is a tool predestined for the diagnosis of prenatal maturation. The analyses over multiple time scales requires sufficiently long data sets while the recordings of fetal heart rate as well as the behavioral states studied are themselves short. Care must be taken that the analysis methods used are appropriate for short data lengths. We investigated multiscale entropy and multifractal scaling exponents from 30 minute recordings of 27 normal fetuses, aged between 23 and 38 weeks of gestational age (WGA) during the quiet state. In multiscale entropy, we found complexity lower than that of non-correlated white noise over all 20 coarse graining time scales investigated. Significant maturation age related complexity increase was strongest expressed at scale 2, both using sample entropy and generalized mutual information as complexity estimates. Multiscale multifractal analysis (MMA) in which the Hurst surface h(q,s) is calculated, where q is the multifractal parameter and s is the scale, was applied to the fetal heart rate data. MMA is a method derived from detrended fluctuation analysis (DFA). We modified the base algorithm of MMA to be applicable for short time series analysis using overlapping data windows and a reduction of the scale range. We looked for such q and s for which the Hurst exponent h(q,s) is most correlated with gestational age. We used this value of the Hurst exponent to predict the gestational age based only on fetal heart rate variability properties. Comparison with the true age of the fetus gave satisfying results (error 2.17±3.29 weeks; p<0.001; R(2)=0.52). In addition, we found that the normally used DFA scale range is non-optimal for fetal age evaluation. We conclude that 30 min recordings are appropriate and sufficient for assessing fetal age by multiscale entropy and multiscale multifractal analysis. The predominant prognostic role of scale 2 heart beats for MSE and scale 39 heart beats (at q=-0.7) for MMA cannot be explored neither by single scale complexity measures nor by standard detrended fluctuation analysis.

Fetal development of complex autonomic control evaluated from multiscale heart rate patterns

Development of the fetal autonomic nervous system's integrative capacity in relation to gestational age and emerging behavioral pattern is reflected in fetal heart rate patterns. Conventional indices of vagal and sympathetic rhythms cannot sufficiently reflect their complex interrelationship. Universal behavioral indices of developing complex systems may provide additional information regarding the maturating complex autonomic control. We investigated fetal magnetocardiographic recordings undertaken at 10-min intervals in active (n = 248) and quiet (n = 111) states between 22 and 39 wk gestational age. Standard deviation of heartbeat intervals, skewness, contribution of particular rhythms to the total power, and multiscale entropy were analyzed. The multiscale entropy methodology was validated for 10-min data sets. Age dependence was analyzed by linear regression. In the quiet state, contribution of sympathovagal rhythms and their complexity over a range of corresponding short scales increased with rising age, and skewness shifted from negative to positive values. In the active state, age dependencies were weaker. Skewness as the strongest parameter shifted in the same direction. Fluctuation amplitude and the complexity of scales associated with sympathovagal rhythms increased. We conclude that in the quiet state, stable complex organized rhythms develop. In the active state, however, increasing behavioral variability due to multiple internal coordinations, such as movement-related heart rate accelerations, and external influences develop. Hence, the state-selective assessment in association with developmental indices used herein may substantially improve evaluation of maturation age and early detection and interpretation of developmental problems in prenatal diagnosis.

Using n-gram analysis to cluster heartbeat signals

Background

Biological signals may carry specific characteristics that reflect basic dynamics of the body. In particular, heart beat signals carry specific signatures that are related to human physiologic mechanisms. In recent years, many researchers have shown that representations which used non-linear symbolic sequences can often reveal much hidden dynamic information. This kind of symbolization proved to be useful for predicting life-threatening cardiac diseases.

Methods

This paper presents an improved method called the “Adaptive Interbeat Interval Analysis (AIIA) method”. The AIIA method uses the Simple K-Means algorithm for symbolization, which offers a new way to represent subtle variations between two interbeat intervals without human intervention. After symbolization, it uses the n-gram algorithm to generate different kinds of symbolic sequences. Each symbolic sequence stands for a variation phase. Finally, the symbolic sequences are categorized by classic classifiers.

Results

In the experiments presented in this paper, AIIA method achieved 91% (3-gram, 26 clusters) accuracy in successfully classifying between the patients with Atrial Fibrillation (AF), Congestive Heart Failure (CHF) and healthy people. It also achieved 87% (3-gram, 26 clusters) accuracy in classifying the patients with apnea.

Conclusions

The two experiments presented in this paper demonstrate that AIIA method can categorize different heart diseases. Both experiments acquired the best category results when using the Bayesian Network. For future work, the concept of the AIIA method can be extended to the categorization of other physiological signals. More features can be added to improve the accuracy.

Glutamate and GABA in the medial amygdala induce selective central sympathetic/parasympathetic cardiovascular responses

Glutamate and γ-aminobutyric acid (GABA) participate in central cardiovascular control, and are found in the rat posterodorsal medial amygdala (MePD), an area of the forebrain that modulates emotional/social behaviors. Here we tested whether these neurotransmitters in the MePD could change the basal activity, chemoreflex, and baroreflex cardiovascular responses in awake rats. Power spectral analysis and symbolic analysis were used to evaluate these responses. Microinjections of saline, glutamate (2 µg), or GABA (61 ng or 100 µg; n = 5–7 rats per group) did not affect basal parameters or chemoreflex responses. However, baroreflex responses showed marked changes. Glutamate increased power spectral and symbolic sympathetic indexes related to both cardiac and vascular modulations (P < 0.05). In turn, the displacement of the baroreflex half-maximal heart rate (HR) response was associated with a GABA (61 ng) mediated decrease in the upper plateau (P < 0.05). Administration of GABA (61 ng, but not 100 µg) also increased HR variability (P < 0.05), in association with parasympathetic activation. These data add novel evidence that the MePD can promote selective responses in the central regulation of the cardiovascular system, i.e., glutamate in the MePD evoked activation of a central sympathetic reflex adjustment, whereas GABA activated a central parasympathetic one.

Effects of coarse-graining on the scaling behavior of long-range correlated and anti-correlated signals

We investigate how various coarse-graining (signal quantization) methods affect the scaling properties of long-range power-law correlated and anti-correlated signals, quantified by the detrended fluctuation analysis. Specifically, for coarse-graining in the magnitude of a signal, we consider (i) the Floor, (ii) the Symmetry and (iii) the Centro-Symmetry coarse-graining methods. We find that for anti-correlated signals coarse-graining in the magnitude leads to a crossover to random behavior at large scales, and that with increasing the width of the coarse-graining partition interval Δ, this crossover moves to intermediate and small scales. In contrast, the scaling of positively correlated signals is less affected by the coarse-graining, with no observable changes when Δ < 1, while for Δ > 1 a crossover appears at small scales and moves to intermediate and large scales with increasing Δ. For very rough coarse-graining (Δ > 3) based on the Floor and Symmetry methods, the position of the crossover stabilizes, in contrast to the Centro-Symmetry method where the crossover continuously moves across scales and leads to a random behavior at all scales; thus indicating a much stronger effect of the Centro-Symmetry compared to the Floor and the Symmetry method. For coarse-graining in time, where data points are averaged in non-overlapping time windows, we find that the scaling for both anti-correlated and positively correlated signals is practically preserved. The results of our simulations are useful for the correct interpretation of the correlation and scaling properties of symbolic sequences.

Binary symbolic dynamics classifies heart rate variability patterns linked to autonomic modulations

Symbolic dynamics derived from heart rate variability (HRV) is able to reflect changes of cardiac autonomic modulations on short time scales in spite of the considerable reduction of information involved. However, the link between the appearance of specific symbolic patterns and the activity of the autonomic nervous system has not yet been elucidated. In this study, we investigate the symbolic dynamics that reflect acceleration (='1') and deceleration (='0') of the instantaneous heart rate. The resulting binary series is analyzed with respect to the regularity of binary patterns of length 8 using Approximate Entropy (ApEn). Binary patterns were grouped according to the level of their regularity as assessed by ApEn. ECG recordings were obtained from 17 healthy subjects during graded head-up tilt (0, 15, 30, 45, 60, 75, and 90°). The linear correlation (Spearman correlation coefficient) between tilt angle and the occurrence of binary patterns was evaluated. The results show that regular binary patterns occurred more often with increasing tilt angle whereas the occurrence of some irregular patterns decreased. Some binary patterns did not show a change of occurrence during tilt. When compared to the results of spectral analysis, regular binary patterns reflect sympathetic modulations whereas irregular binary patterns reflect parasympathetic modulations. The parameters derived from binary symbolic dynamics reflect changes of autonomic modulations during graded head-up tilt and are not fully correlated to the spectral markers of HRV.

Multiscale analysis of acceleration and deceleration of the instantaneous heart rate using symbolic dynamics

The multiscale analysis of physiologic time series such as the RR interval time series has revealed that the entropy differs according to the scale. Furthermore, healthy subjects show different characteristics on the different time scales compared to patients. Instead of calculating entropies of the time series, the sequence of acceleration and deceleration of the instantaneous heart rate may also be investigated by means of binary symbolic dynamics. This kind of analysis revealed that the healthy heartbeat series also contains numerous regular binary sequences indicating runs of acceleration or deceleration. Here, we investigate whether this approach yields new information when applied to multiple time scales. We investigate the occurrence of binary patterns of length 8 on different time scales of heart rate series from healthy subjects and patients with congestive heart failure (CHF). Healthy subjects and CHF patients show different occurrences of binary patterns. These occurrences change especially on scales 1 to 5. Healthy subjects show more pronounced changes than CHF patients. At larger scales only gradual changes were observed. In conclusion, the application of binary symbolic dynamics on different scales yields new information, in particular on small scales.

Autonomic cardiovascular modulation

We validated a symbolic approach to assess autonomic modulation from pulse interval (PI) and systolic arterial pressure (SAP) series obtained from an animal model of chronic heart failure (CHF). We studied three groups of rats: controls; CHF animals; CHF animals treated with spironolactone (CHF-SP), reducing sympathetic activity in CHF. Simulations confirmed that symbolic analysis captures modifications of cardiovascular regulation in the case of fast dynamics and negligible variance. While spectral indexes did not reveal any significant difference among groups, symbolic analysis pointed out that sympathetic modulation is reduced in CHF group and restored to basal values in CHF-SP one.

Testing foetal-maternal heart rate synchronization via model-based analyses

The investigation of foetal reaction to internal and external conditions and stimuli is an important tool in the characterization of the developing neural integration of the foetus. An interesting example of this is the study of the interrelationship between the foetal and the maternal heart rate. Recent studies have shown a certain likelihood of occasional heart rate synchronization between mother and foetus. In the case of respiratory-induced heart rate changes, the comparison with maternal surrogates suggests that the evidence for detected synchronization is largely statistical and does not result from physiological interaction. Rather, they simply reflect a stochastic, temporary stability of two independent oscillators with time-variant frequencies. We reanalysed three datasets from that study for a more local consideration. Epochs of assumed synchronization associated with short-term regulation of the foetal heart rate were selected and compared with synchronization resulting from white noise instead of the foetal signal. Using data-driven modelling analysis, it was possible to identify the consistent influence of the heartbeat duration of maternal beats preceding the foetal beats during epochs of synchronization. These maternal beats occurred approximately one maternal respiratory cycle prior to the affected foetal beat. A similar effect could not be found in the epochs without synchronization. Simulations based on the fitted models led to a higher likelihood of synchronization in the data segments with assumed foetal-maternal interaction than in the segment without such assumed interaction. We conclude that the data-driven model-based analysis can be a useful tool for the identification of synchronization.

Nonlinear properties of vagal and sympathetic modulations of heart rate variability in ovine fetus near term

Fetal heart rate (FHR) monitoring is commonly used although clinical studies questioned its diagnostic value. Sophisticated FHR variability (fHRV) measures such as fHRV complexity may improve the sensitivity and specificity of FHR monitoring. A more detailed understanding of the physiology underlying fHRV complexity is essential to harness its use for monitoring fetal health. To examine the specific effects of vagal and sympathetic modulations on fHRV complexity, we blocked vagal activity with atropine and sympathetic activity with propranolol in near-term fetal sheep ( n = 7, 0.85 gestation). Under these conditions, we analyzed the linear and nonlinear parts of fHRV complexity from autonomic information flow. Overall fHRV complexity decreased with both drugs compared with nonrapid eye movement sleep baseline ( P < 0.05). With atropine, this was because of a decrease of the linear part of fHRV complexity on the long-term time scale ( P < 0.05), suggesting that vagal modulation of fHRV is adequately described by linear fHRV measures. With propranolol, the nonlinear part of fHRV complexity decreased on the short-term time scale ( P < 0.05), suggesting that sympathetic influences on fHRV can be detected by the nonlinear part of fHRV complexity. Thus the complex interplay of vagal and sympathetic modulations of fHRV is reflected differently and specifically in the linear and nonlinear properties of fHRV complexity, and on different time scales. Analysis of linear and nonlinear properties of fHRV may improve sensitivity and specificity of FHR monitoring.

Heart rate features in fetal behavioural states

BACKGROUND AND AIM

Fetal behavioural states have been defined on the basis of eye movements, body movements and heart rate patterns as presented by cardiotocography (CTG). The aim of this work was to determine whether behavioural states can be distinguished on the basis of heart rate features alone using high resolution beat-to-beat fetal magnetocardiography.

STUDY DESIGN

Five minute magnetocardiograms were recorded at a sampling rate of 1 kHz in 40 healthy fetuses (36th-41st week of gestation). In the reconstructed RR interval time series, 256-beat epochs corresponding to the behavioural states 1F, 2F and 4F were visually identified according to heart rate patterns as defined for CTG. These epochs were then quantified using mean RR interval, its standard deviation (SDNN), its root mean square of successive difference (RMSSD) and on the basis of symbolic dynamics of short 8 beat trains.

RESULTS

Pairwise comparison between the behavioural states showed that the values of each of these measures differed significantly between the states. Quadratic discriminant analysis further revealed that mean RR interval and SDNN sufficed to classify state with a correct classification of 92%.

CONCLUSIONS

The results suggest that measures that quantify overall aspects of heart rate can distinguish RR interval time series which were classified into different fetal behavioural states. The differences in short-term variability as quantified by RMSSD and symbolic dynamics may help reveal new aspects of these states.

Introduction: Cardiovascular physics

The number of patients suffering from cardiovascular diseases increases unproportionally high with the increase of the human population and aging, leading to very high expenses in the public health system. Therefore, the challenge of cardiovascular physics is to develop high-sophisticated methods which are able to, on the one hand, supplement and replace expensive medical devices and, on the other hand, improve the medical diagnostics with decreasing the patient's risk. Cardiovascular physics-which interconnects medicine, physics, biology, engineering, and mathematics-is based on interdisciplinary collaboration of specialists from the above scientific fields and attempts to gain deeper insights into pathophysiology and treatment options. This paper summarizes advances in cardiovascular physics with emphasis on a workshop held in Bad Honnef, Germany, in May 2005. The meeting attracted an interdisciplinary audience and led to a number of papers covering the main research fields of cardiovascular physics, including data analysis, modeling, and medical application. The variety of problems addressed by this issue underlines the complexity of the cardiovascular system. It could be demonstrated in this Focus Issue, that data analyses and modeling methods from cardiovascular physics have the ability to lead to significant improvements in different medical fields. Consequently, this Focus Issue of Chaos is a status report that may invite all interested readers to join the community and find competent discussion and cooperation partners.