The Autonomic System Functional State Predicts Responsiveness in Disorder of Consciousness

Abstract

Diagnosis and early prognosis of the vegetative state/unresponsive wakefulness syndrome (VS/UWS) and its differentiation from the minimally-conscious state still rest on the clinical observation of responsiveness. The incidence of established clinical indicators of responsiveness also has proven variable in the single subject and is correlated to measures of heart rate variability (HRV) describing the sympathetic/parasympathetic balance. We tested responsiveness when the HRV descriptors nuLF and peakLF were or were not in the ranges with highest incidence of response based on findings from previous studies (10.0–70.0 and 0.05–0.11 Hz, respectively). Testing was blind by The Coma Recovery Scale-revised in the two conditions and in two experimental sessions with a one-week interval. The incidence of responses was not randomly distributed in the “response” and “no-response” conditions (McNemar test; p<0.0001). The observed incidence in the “response” condition (visual: 55.1%; auditory: 51.5%) was higher than predicted statistically (32.1%) or described in previous clinical studies; responses were only occasional in the “no-response” condition (visual, 15.9%; auditory, 13.4%). Models validated the predictability with high accuracy. The current clinical criteria for diagnosis and prognosis based on neurological signs should be reconsidered, including variability over time and the autonomic system functional state, which could also qualify per se as an independent indicator for diagnosis and prognosis.

Introduction

The Vegetative State (or Unresponsive Wakefulness Syndrome; VS/UWS)¹ is characterized by preserved arousal in the absence of awareness and communication in the severely brain damaged patient^1

–7; (fragmentary) behavioral patterns compatible with awareness mark evolution into the minimally-conscious state (MCS).^8,9 Neuroimaging studies have documented in VS/UWS regional brain activation in response to stimulus conditions at varying level of complexity—from phase coding to pain to sensory/cognitive processing to emotional empathy or reactivity to learning.^{3,10

–15} These observations allow speculation that all subjects with brain activation documented by functional imaging may be in a MCS.¹² However, functional imaging is only available in selected centers and at present is mainly a research tool; diagnosis and early prognosis of VS/UWS and differentiation from MCS still rest today mainly on the clinical observation of responsiveness.^12,14,16,17

The incidence of established clinical indicators of responsiveness has proven variable also in the single subject. The Coma Recovery Scale-revised (CRS-r)^9,16 global, visual, and auditory scores were found to be higher in the morning than in the afternoon in subjects repeatedly tested.¹⁸ The visual pursuit response (a key CRS-r item and marker of evolution into the MCS)^9,16 reportedly is observed in about 20% of subjects otherwise unambiguously diagnosed as being in VS/UWS and in about 70% of cases in MCS.¹⁹ Its incidence at multiple testing (six times/subject/d) varied within/across subjects and over time during the day, with maximal incidence occurring at 10.30 am and 3.00 pm and overall chances of observing it at least once during the day of about 33% and 62% in VS/UWS and MCS patients, respectively.²⁰ The pursuit response incidence in these conditions and its variability during the day are of the same magnitude of the estimated misdiagnosis rate between VS/UWS and MCS (up to 25–45%).^{7,12,21
–23}

Replicable changes in the autonomic system (ANS) functional state have been identified in VS/UWS and MCS subjects in response to simple or complex sensory inputs.^24

–28 The pursuit response proved correlated to measures of heart rate variability (HRV; the heart rate fluctuations around the mean value over the time sample) describing the sympathetic/parasympathetic balance. The response was modeled by machine learning methodologies (Support Vector Machine)²⁹ and the procedure confirmed the correlation. The intervals of HRV descriptors at which the incidence of the visual response was maximal were identified.³² Contributions by the ANS in modulating the residual responsiveness would be in agreement with its predominant role in VS/UWS subjects.^{3,10,30
–32} The purpose of this study was to verify whether the HRV descriptors of autonomic function can predict the incidence of responses in VS/UWS subjects with an accuracy compatible with clinical needs. To this end, we measured HRV, an established index of the sympathetic/parasympathetic interplay also applied in the description of interactions between the brain and the ANS both in awake subjects and those with a severe disorder of consciousness.^33

–37 Two separate conditions were selected (i.e., when the values of the HRV significant markers were or were not within the ranges at which responsiveness was observed with higher incidence in previous studies).^27,38

–41 Each subject was blindly tested for response in each condition.

Methods

Patients

Eight subjects (seven males, age-range 18–55 years; mean age 39.6±17.3 years; one female, age 23 years) in chronic VS/UWS according to the current clinical criteria and established evaluation scales were studied.^2,4
–6,8 In all cases, scores were congruent with the cutoffs in use for clinical characterization of the subjects in VS/UWS—that is, lower than 25 on the Loewenstein Scale⁴² and 2 on the Level of Cognitive Function scale (LCF),⁴³ and higher than 21 on the Disability Rating Scale.⁴⁴ The CRS-r¹⁶ global score was equal to or lower than 7. Subjects who were clinically unstable, had concurrent systemic disorders (e.g., of the cardiovascular system), were under treatment with (neuro)active drugs or beta-blockers, or had evidence of recurrent pain were not admitted to the study. The patients' relatives or caregivers were informed in full detail about the study and procedures and gave their written consent. Data were treated under the condition of anonymity. The ethical principles of the Declaration of Helsinki (1964) by the World Medical Association concerning human experimentation were followed. Demographics and clinical information are summarized in Table 1.

Table 1.

Demographics and Clinical Information

	Sex	Age	Etiology	CRS-r	Days in VS/UWS
1	m	53	Ischemic	6	67
2	m	41	Traumatic	6	70
3	m	51	Traumatic	4	78
4	m	55	Hemorrhagic	6	60
5	f	23	Traumatic	4	75
6	m	18	Traumatic	6	71
7	m	18	Traumatic	7	82
8	m	58	Hemorrhagic	4	66

CRS-r, Coma Recovery Scale-revised; VS/UWS, vegetative state/unresponsive wakefulness syndrome; m, male; f, female.

Stimulus conditions and experimental procedure

Subjects were tested between 10:00 am and 11:00 am, never sooner than 30–45 min after nursing care, and always before rehabilitation or other treatments.²⁰ They were comfortably lying on armchairs with constant 24°C ambient temperature, in absence of transient noises or environmental interferences. When tested, subjects had their eyes open and no sign of sleep or drowsiness could be detected upon observation; in order to minimize uncontrolled interference and individual variability, no sensory or noxious stimuli or Arousal Facilitation Protocols (AFP) favoring wakefulness or arousal were administered prior to testing.

Each patient was tested twice with a week interval in recording sessions lasting 20±1 min. Each session included 24 stimulus conditions (12 visual, 12 auditory) presented according to the CRS-r guidelines.¹⁶ Specifically, the visual pursuit response was tested by means of a round mirror (12 cm in diameter) moved slowly in front of the subject in the horizontal and vertical planes for 45° (right, left, up, down); it was ranked positive for a consistent response when the patient's eyes followed the mirror without loss of fixation. Responsiveness to auditory stimuli was tested by calling the patient's name from the right or left side while standing behind him/her to avoid visual contact; the patient's moving the head toward the calling voice was regarded as a positive response.¹⁶

Previous studies applying advanced data analyses to sort out consistent trends or associations in large datasets⁴⁵ have identified the HRV descriptors nuLF and pkLF as reliable measures of sympathetic/parasympathetic function in the VS/UWS and MCS.^{34,38,46
–48} Both parameters describe the control of cardiovascular function in adaptive behavior and the interaction between excitatory and inhibitory autonomic control mechanisms. Specifically, nuLF is deemed indicative of sympathovagal balance.⁴⁹ The 0.1 Hz component LF (pkLF) describes the sympathetic/parasympathetic interplay mediated by baroreflex and has been related to changes in arousal and to emotional responses.^50,51 A correlation between responsiveness and these descriptors is documented.^26,39,41 The electrocardiogram (EKG) was continuously recorded at rest (5 min baseline) and during testing, and the HRV descriptors were computed online. Visual and auditory stimuli were administered only when the HRV signal was stationary in order to reduce short fluctuations. Two testing conditions were predetermined according to previous observations and models⁴¹ indicating a highest probability of observing a response at HRV nuLF values in the 10–70 interval and peakLF values between 0.05 Hz and 0.11 Hz; these intervals are defined (and will hereafter be referred to) as the “response” condition (Fig. 1 and Fig. 2). Any time point with nuLF and peakLF values outside these ranges was considered a “no-response” condition.

FIG. 1.

Bottom: In vertical bars, the numbers of observed (left; n=71) and not observed (right; n=150) responses to sensory stimulation. In green, on the horizontal planes, is the “response” condition as pre-defined by the heart rate variability (HRV) nuLF and peakLF descriptors values at which the highest incidence of responses has been observed in previous studies; in red, the “no-response” HRV condition. Top: Three-dimensional distribution of the probability of observing or not observing a response, estimated as the relative frequency of response for each subject versus the HRV descriptors.

FIG. 2.

Numbers of observed responses to visual (right, n=36) and auditory (left, n=35) stimulus conditions and distributions of the probability of observing a response estimated as the relative frequency of response for each subject versus the heart rate variability descriptors. In green and red, the “response” and the “no response” areas, respectively.

Each stimulus was administered twice by an expert neuropsychologist, who was blind for the “response” or “no-response” condition of the HRV parameters and ANS functional state. A total of 384 stimuli were administered; 42% were discarded because of artifacts due to movement; 221 stimuli (27.6±4.3 per subject) were analyzed.

HRV data collection

The heart rate signal was recorded through a photopletismograph sensor positioned on the middle finger of the left hand; the sampling rate was 128 Hz, with resolution at 24 bit (NEXUS 32; Mindmedia, Herten, the Netherlands).^52
–54 This procedure was favored to minimize the patients' discomfort due to conventional EKG recording electrodes; recording procedure and sampling rate have proved reliable in previous studies on awake^55,56 and VS/UWS subjects^24,27,28,41 and are in agreement with the guidelines by the Task Force of European Society of Cardiology and the North American Society of Pacing and Electrophysiology.⁵⁷ The tachogram (i.e., the series of consecutive intervals between heartbeats) was processed in the frequency domain (Fast Fourier Transform [FFT]) by the software Biotratace+ (www.mindmedia.nl) developed for physiological monitoring, data analysis, signal processing, and clinical biofeedback-neurofeedback.^58

–61 FFT was performed in the 0.01 to 0.5 Hz interval with 0.001 Hz resolution and the spectral descriptors in three frequency ranges (very low frequency, 0.01–0.04 Hz; low frequency [LF], 0.04–0.15 Hz; and high frequency, 0.15–0.5 Hz) were computed and processed statistically. The normalized LF index nuLF⁴⁹ and the peakLF^40,50,62 were computed in real time by mean Nexus 32 and Biotrace software.

The signal stationarity in the time windows at which responsiveness to visual or auditory stimulus conditions was or was not to be tested (the “response” and “no-response” conditions) was evaluated in real time by means of the graphical representation of both the row signals and of the coefficient of variability calculated in 15-sec time windows.

Data processing

The baseline HRV descriptors of the two recording sessions were compared using the Mann-Whitney test and data were cumulated for successive analyses after excluding differences. The “response” and “no-response” conditions were compared for the probability of observing a visual or auditory response by the McNemar's test^63,64 after estimating the expected distribution of responses in the two conditions by the Bayesian Analysis Model⁶⁵ (pre-test [prevalence] and post-test [positive predictive values] prevalence). The incidences of responses in the “response” and “no-response” conditions and in the first and second sessions were compared using the Fisher's exact test.

The machine learning classifier Support ONE-R was applied to estimate the weight of each selected HRV descriptor (nuLF and pkLF).⁶⁶ ONE-R is a simple classification algorithm that generates one level of decisional tree and infers simple classification rules from a set of instances; in this case, the aim was to model a rule predicting the class (response, no-response) given the values of the attributes (HRV descriptors).^66,67 The algorithm ONE-R is commonly used to obtain a good trade-off between recognition (portion of correct sample/label association in a training sub-sample) and generalization (portion of correct sample/label association in a test sub-sample). The 10 fold cross validation statistics were used to partition the data samples into subsets so that data analysis was first carried on selected subsets, while other subsets served to confirm and validate the initial analyses.⁶⁸

Results

No differences were found when comparing in the frequency domain the HRV descriptors from the two baseline sessions at a one-week interval (Mann-Whitney test: −1.050<z < −0.053; p>0.29).

The Bayesian analysis model predicted an incidence of 32.1% observed responses in the pre-test (prevalence) and 52.8% in the post-test (positive predictive values) analyses, with a negative predictive value of 86.9% and positive and negative likelihood ratios of 2.3662 and 0.3169, respectively. Positive responses were observed irrespective of the sensory modality in the 53.3% and 12.9% of cases in the “response” and “no-response” conditions, respectively; the mean incidence per subject of positive responses was 53.1±21.1 and 14.6±10.3 in the two conditions, respectively. The observed responses were not randomly distributed but were clustered in each session at the selected nuLF values between 10 and 30 and peakLF values between 0.06 Hz and 0.11 Hz. The differences in distribution were significant on the McNemar test (X2=17.455; p<0.0001), without substantial differences between the visual and auditory stimulus conditions (Fig. 1 and Fig. 2).

The incidence of observed visual or auditory responses in the “response” condition was 55.1% (72% at nuLF values between 10 and 30) and 51.7%, respectively (mean per subject, 54.3±32.2 [visual] and 51.9±19.7 [auditory]); that of no observable response in the “no-response” condition was 83.3% (visual) and 90% (auditory; Fig. 3). No differences were found between the two testing sessions in the occurrence (Fisher's exact test: p=0.43) or absence (p=0.57) of responses in the “response” and “no-response” conditions.

FIG. 3.

Overall incidence of the observed responses in the “response” and “no-response” conditions.

The ONE-R classifier estimated a correct classification of 76% and 72.4% in the training test and of 66.9% and 69.2% in the 10-fold cross-validation test for the nuLF and peakLF descriptors, respectively (Table 2). The correct prediction of response by the nuLF descriptor (43.7% positive predictive value) was higher than by the peakLF descriptor (24.6%). The highest correct classification of responses was observed at nuLF values between 21 and 30, and peakLF values between 0.09 HZ and 0.11Hz. Specificity was high (71.81–80.89).

Table 2.

Classification by the ONE-R Classifier

	Training test		10-Fold cross validation test
	nuLF	peakLF	nuLF	peakLF
Correct classification	76.0	72.4	66.9	69.23
Sensitivity	64.06	64.71	48.44	54.55
specify	80.89	73.37	74.52	71.81
Positive predictive values (response)	57.75	30.99	43.66	25.35
Negative predictive values (no-response)	84.67	91.84	78	90
Fisher's exact test (p)	0.0001	0.0001	0.0013	0.004

Discussion

Evidence from neuroimaging research has documented retained modular brain function in VS/UWS and MCS subjects in the absence of the integrative processes necessary to sustain consciousness.^{3,11,13,14,69} Neuroimaging studies also indicate direct/indirect functional links between the autonomic control and activity in brain structures that are involved in higher brain functions, such as attention and conscious processes.^{33,35,36,70,71} A model network has been proposed, with the anterior cingulate cortex and its projections to the prefrontal cortex, amigdala, hypothalamus and brainsteam involved in the modulation of autonomic output in response to pain and emotional or behavioral stimulus conditions^{33,37,72

–75}; the model is consistent with the stimulus-related changes in the autonomic functional state observed in subjects otherwise diagnosed as being in VS/UWS.^{24,26
–28,41} Responsiveness in these subjects, therefore, appears to be mediated by relatively complex processes. The incidence of established clinical signs of responsiveness varied during the day^18,20 and has been found to correlate with the sympathetic/parasympathetic balance.⁴¹

In our study, responsiveness was predicted with high accuracy by the sympathetic/parasympathetic balance measured by the HRV descriptors nuLF and pkLF. The observation of visual (54.8%) or auditory (55.3%) responses was significantly higher in our subjects' group when tested at nuLF and pkLF descriptors within the value ranges indicative of sympathetic/parasympathetic balance than outside these ranges (12.9% and 16%, respectively). The incidence of a visual pursuit response was about twice higher at these HRV values than when predicted statistically or compared with the reported incidence at random testing (∼20%).^6,19 Remarkably, the absence of responses was predicted with over 85% accuracy when testing the subject's responsiveness at HRV values outside the “response” intervals but only in about half of cases when testing within the “response” intervals. Taken together, these findings suggest that subjects in VS/UWS may differ as to responsiveness to conventional sensory stimuli and could be classified in this respect with greater accuracy if tested in relation to their autonomic system functional state.

The test–retest experimental paradigm has been designed and limited in time to minimize the effect of age on HRV^41,76,77 and those of fragmentary circadian/ultradian or otherwise cyclic processes, spontaneous fluctuations, and environmental interference, as well as the end-effects of neuronal/nonneuronal factors modulating the brain state or responsiveness.^{3,20,78
–80} Sources of biases possibly gone undetected in the study and random variability cannot be ruled out in principle but the higher incidence of responses at pre-selected ranges of the HRV descriptors is likely to exceed spontaneous variability and rather suggests that the current indices of responsiveness in VS/UWS vary depending on the concomitant functional state/balance of the autonomic nervous system.^{26,27,38,41,81} In particular, some pathophysiological specificity of the HRV measures appears possible and these findings confirm the CRS-r reliability in the identification of conditions of different underlying pathophysiology.

Diagnosis and the clinical differentiation between VS/UWS and MCS remain mandatory in the management of patients, allocation of resources, and programming of health care policies but still rest mainly on the clinical observation of responsiveness and on the clinician's responsibility,^9,12,23 with a high estimated misdiagnosis between VS/UWS and MCS.^{12,19,21
–23} Contrary to the positron emission tomography and functional magnetic resonance imaging availability mainly as research tools,¹² HRV recording methods are noninvasive and of limited technical complexity, easily applicable in conditions allowing limited environmental control such as (semi)intensive care units or with no collaboration by the investigated subjects, as it is in the case in the disorder of consciousness; thus, standard HRV analyses can be easily implemented.^{34,47,82
–84} Testing for responsiveness in the disorder of consciousness while monitoring the ANS functional state and balance would result in higher accuracy and seems a practical approach to control the risks of erroneous classification and misdiagnosis of any individual subject in VS/UWS. Validation by mathematical models⁴¹ allows some generalization of results due to the low sensitivity to risks of alpha inflation⁸⁵ also when derived from a relatively large dataset (n=221 in our case) from a small subject sample.^66,68,86,87 Caution is nevertheless mandatory and the translational power of the approach to the clinical routine must be confirmed in cohort investigations on large patients' samples.

It has been suggested that VS/UWS appears today neither static nor homogeneous, with characteristics compatible with a multifaceted condition—or variety of conditions.⁷ Accordingly, the clinical criteria currently in use to define the disorders of consciousness and predict outcome based on neurological signs should be reconsidered, including variability over time and correlation with the autonomic system functional state—which could also qualify per se as an independent variable for diagnosis and prognosis.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Laureys

, Celesia

G.G.

, Cohadon

, Lavrijsen

, León-Carrión

, Sannita

W.G.

, Sazbon

, Schmutzhard

, Wild

K.R.

, von Zeman

, and Dolce

(2010). Unresponsive wakefulness syndrome: a new name for the vegetative state or apallic syndrome. BMC Med., 8, 68.

The Multi-Society Task Force on PVS. (1994). Medical aspects of the persistent vegetative state. N Engl J Med, 330, 1499–1508, 1572.

Owen

A.M.

, Coleman

M.R.

, Boly

, Davis

M.H.

, Laureys

, and Pickard

J.D.

(2006). Detecting awareness in the vegetative state. Science, 313, 1402–1402.

Monti

M.M.

, Laureys

, and Owen

A.M.

(2010). The vegetative state. BMJ, 341, c3765–c3765.

Jennett

(2002). The vegetative state. J. Neurol. Neurosurg. Psychiatry, 73, 355–357.

Dolce

and Sazbon

(2002). The Post-Traumatic Vegetative State. Thieme Medical Publishers: Stuttgart, Germany.

Dolce

and Sannita

W.G.

; European Task Force On The Vegetative State. (2010). The vegetative state: a syndrome seeking revision?. Brain Inj., 24, 1628–1629.

Giacino

J.T.

(1997). Disorders of consciousness: differential diagnosis and neuropathologic features. Semin. Neurol., 17, 105–111.

Giacino

J.T.

, Ashwal

, Childs

, Cranford

, Jennett

, Katz

D.I.

, Kelly

J.P.

, Rosenberg

J.H.

, Whyte

, Zafonte

R.D.

, and Zasler

N.D.

(2002). The minimally conscious state: definition and diagnostic criteria. Neurology, 58, 349–353.

10.

Riganello

, and Sannita

W.G.

(2009). Residual brain processing in the vegetative state. J. Psychophysiol., 23, 18–26.

11.

Bekinschtein

, Niklison

, Sigman

, Manes

, Leiguarda

, Armony

, Owen

, Carpintiero

, and Olmos

(2004). Emotion processing in the minimally conscious state. J. Neurol. Neurosurg. Psychiatry, 75, 788–788.

12.

Celesia

G.G.

, and Sannita

W.G.

(2013). Can patients in vegetative state experience pain and have conscious awareness?. Neurology, 80, 328–329.

13.

Laureys

, Faymonville

M.E.

, Peigneux

, Damas

, Lambermont

, Del

Fiore

, G., Degueldre

, Aerts

, Luxen

, Franck

, Lamy

, Moonen

, and Maquet

(2002). Cortical processing of noxious somatosensory stimuli in the persistent vegetative state. NeuroImage, 17, 732–741.

14.

Monti

M.M.

(2012). Cognition in the Vegetative State. Annu. Rev. Clin. Psychol., 8, 431–454.

15.

Sharon

, Pasternak

, Ben

Simon

, E., Gruberger

, Giladi

, Krimchanski

B.Z.

, Hassin

, and Hendler

(2013). Emotional Processing of Personally Familiar Faces in the Vegetative State. PLoS ONE, 8, e74711.

16.

Giacino

J.T.

, Kalmar

, and Whyte

(2004). The JFK Coma Recovery Scale-Revised: Measurement characteristics and diagnostic utility. Arch. Phys. Med. Rehabil., 85, 2020–2029.

17.

Vanhaudenhuyse

, Schnakers

, Brédart

, and Laureys

(2008). Assessment of visual pursuit in post-comatose states: use a mirror. J. Neurol. Neurosurg. Psychiatry, 79, 223–223.

18.

Cortese

M.D.

, Riganello

, Arcuri

, Pugliese

M.E.

, Lucca

L.F.

, Giuliano

, and Sannita

W.G.

(2014). Coma Recovery Scale-r: variability in the disorder of consciousness. BMC Neurol. Submitted.

19.

Dolce

, Lucca

L.F.

, Candelieri

, Rogano

, Pignolo

, and Sannita

W.G.

(2010). Visual pursuit in the severe disorder of consciousness. J. Neurotrauma, 28, 1149–1154.

20.

Candelieri

, Cortese

M.D.

, Dolce

, Riganello

, and Sannita

W.G.

(2011). Visual pursuit: within-day variability in the severe disorder of consciousness. J. Neurotrauma, 28, 2013–2017.

21.

Andrews

, Murphy

, Munday

, and Littlewood

(1996). Misdiagnosis of the vegetative state: retrospective study in a rehabilitation unit. BMJ, 313, 13–16.

22.

Bosco

, Lancioni

G.E.

, Belardinelli

M.O.

, Singh

N.N.

, O'Reilly

M.F.

, and Sigafoos

(2010). Vegetative state: efforts to curb misdiagnosis. Cogn. Process., 11, 87–90.

23.

Schnakers

, Vanhaudenhuyse

, Giacino

, Ventura

, Boly

, Majerus

, Moonen

, and Laureys

(2009). Diagnostic accuracy of the vegetative and minimally conscious state: Clinical consensus versus standardized neurobehavioral assessment. BMC Neurol., 9, 35.

24.

Riganello

, Quintieri

, Candelieri

, Conforti

, and Dolce

(2008). Heart rate response to music: An artificial intelligence study on healthy and traumatic brain-injured subjects. J. Psychophysiol., 22, 166–174.

25.

Machado

, Korein

, Aubert

, Bosch

, Alvarez

M.A.

, Rodríguez

, Valdés

, Portela

, García

, Pérez

, Chinchilla

, Machado

, and Machado

(2007). Recognizing a mother's voice in the persistent vegetative state. Clin. EEG Neurosci., 38, 124–126.

26.

Gutiérrez

, Machado

, Estévez

, Olivares

, Hernández

, Perez

, Beltrán

, and Leisman

(2010). Heart rate variability changes induced by auditory stimulation in persistent vegetative state. Int. J. Disabil. Hum. Dev., 9, 357–362.

27.

Riganello

, Candelieri

, Quintieri

, Conforti

, and Dolce

(2010). Heart rate variability: An index of brain processing in vegetative state?. An artificial, intelligence, data mining study. Clin. Neurophysiol. 121, 2024–2034.

28.

Riganello

, Candelieri

, Quintieri

, and Dolce

(2010). Heart rate variability, emotions, and music. J. Psychophysiol., 24, 112–119.

29.

Noble

W.S.

(2006). What is a support vector machine?. Nat. Biotechnol., 24, 1565–1567.

30.

Kotchoubey

, Lang

, Mezger

, Schmalohr

, Schneck

, Semmler

, Bostanov

, and Birbaumer

(2005). Information processing in severe disorders of consciousness: Vegetative state and minimally conscious state. Clin. Neurophysiol., 116, 2441–2453.

31.

, Boly

, Weng

, Ledoux

, and Laureys

(2008). Neuroimaging activation studies in the vegetative state: predictors of recovery?. Clin. Med., 8, 502–507.

32.

Boly

, Tshibanda

, Vanhaudenhuyse

, Noirhomme

, Schnakers

, Ledoux

, Boveroux

, Garweg

, Lambermont

, Phillips

, Luxen

, Moonen

, Bassetti

, Maquet

, and Laureys

(2009). Functional connectivity in the default network during resting state is preserved in a vegetative but not in a brain dead patient. Hum. Brain Mapp., 30, 2393–2400.

33.

Thayer

J.F.

, and Lane

R.D.

(2009). Claude Bernard and the heart–brain connection: further elaboration of a model of neurovisceral integration. Neurosci. Biobehav. Rev., 33, 81–88.

34.

Riganello

, Garbarino

, and Sannita

W.G.

(2012). Heart rate variability, homeostasis, and brain function: a tutorial and review of application. J. Psychophysiol., 26, 178–203.

35.

Napadow

, Dhond

, Conti

, Makris

, Brown

E.N.

, and Barbieri

(2008). Brain correlates of autonomic modulation: Combining heart rate variability with fMRI. NeuroImage, 42, 169–177.

36.

Lane

R.D.

, McRae

, Reiman

E.M.

, Chen

, Ahern

G.L.

, and Thayer

J.F.

(2009). Neural correlates of heart rate variability during emotion. NeuroImage, 44, 213–222.

37.

Riganello

, Dolce

, and Sannita

(2012). Heart rate variability and the central autonomic network in the severe disorder of consciousness. J. Rehabil. Med., 44, 495–501.

38.

Wijnen

V.J.M.

, Heutink

, van Boxtel

G.J.M.

, Eilander

H.J.

, and Gelder

B. de.

(2006). Autonomic reactivity to sensory stimulation is related to consciousness level after severe traumatic brain injury. Clin. Neurophysiol., 117, 1794–1807.

39.

Riganello

, Candelieri

, Dolce

, and Sannita

W.G.

(2011). Residual emotional processing in the vegetative state: A scientific issue?. Clin. Neurophysiol., 122, 1061–1062.

40.

Nickel

, and Nachreiner

(2003). Sensitivity and diagnosticity of the 0.1-Hz component of heart rate variability as an indicator of mental workload. Hum. Factors J. Hum. Factors Ergon. Soc., 45, 575–590.

41.

Riganello

, Cortese

M.D.

, Dolce

, and Sannita

W.G.

(2013). Visual pursuit response in the severe disorder of consciousness: modulation by the central autonomic system and a predictive model. BMC Neurol., 13, 164.

42.

Borer-Alafi

, Gil

, Sazbon

, and Korn

(2002). Loewenstein communication scale for the minimally responsive patient. Brain Inj., 16, 593–609.

43.

Hagen

, Malkmus

, and Durham

(1972). Rancho Los Amigos Levels of Cognitive Functioning Scale. Downey CA Prof. Staff Association: Downey California.

44.

Rappaport

, Hall

Km.

, Hopkins

, Belleza

, and Cope

Dn.

(1982). Disability rating scale for severe head trauma: coma to community. Arch. Phys. Med. Rehabil., 63, 118–123.

45.

Candelieri

, Dolce

, Riganello

, and G

(2011). Data mining in neurology. In: Knowledge-Oriented Applications in Data Mining. Funatsu

. (ed). InTech: Rijeka, Croatia.

46.

Candelieri

, Riganello

, Cortese

, and Sannita

W.G.

(2011). Functional status and the eye-tracking response—A data mining classification study in the vegetative and minimally conscious states. HEALTHINF 2011—Proceedings of the International Conference on Health Informatics, Rome, Italy.

47.

Ryan

M.L.

, Thorson

C.M.

, Otero

C.A.

, Vu

, and Proctor

K.G.

(2011). Clinical applications of heart rate variability in the triage and assessment of traumatically injured patients. Anesthesiol. Res. Pract., 2011, e416590.

48.

Wieser

, Buetler

, Koenig

, and Riener

(2010). Quantitative description of the state of awareness of patients in vegetative and minimally conscious state. 2010 Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC).

49.

Burr

R.L.

(2007). Interpretation of normalized spectral heart rate variability indices in sleep research: a critical review. Sleep, 30, 913–919.

50.

Nickel

, and Nachreiner

(2000). Psychometric Properties of the 0.1 HZ Component of HRV as an indicator of mental strain. Proc. Hum. Factors Ergon. Soc. Annu. Meet., 44, 2–747–2–750.

51.

Nickel

, Nachreiner

, and Yanagibori

(2003). The 0.1 Hz component of HRV as an indicator of mental workload: Sensitivity for intensity and time on task effects. Int. J. Psychophysiol., 17, 94.

52.

Bolanos

, Nazeran

, and Haltiwanger

(2006). Comparison of Heart Rate Variability Signal Features Derived from Electrocardiography and Photoplethysmography in Healthy Individuals. In: 28th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2006. Engineering in Medicine and Biology Society, pps. 4289–4294.

53.

Selvaraj

, Jaryal

, Santhosh

, Deepak

K.K.

, and Anand

(2008). Assessment of heart rate variability derived from finger-tip photoplethysmography as compared to electrocardiography. J. Med. Eng. Technol., 32, 479–484.

54.

, Yang

, Taylor

J.A.

, and Stein

J.F.

(2009). A comparison of photoplethysmography and ECG recording to analyse heart rate variability in healthy subjects. J. Med. Eng. Technol., 33, 634–641.

55.

Gil

, Orini

, Bailon

, Vergara

J.M.

, Mainardi

, and Laguna

(2010). Time-varying spectral analysis for comparison of HRV and PPG variability during tilt table test., in: 2010 Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). pps. 3579–3582.

56.

Gil

, Orini

, Bailón

, Vergara

J.M.

, Mainardi

, and Laguna

(2010). Photoplethysmography pulse rate variability as a surrogate measurement of heart rate variability during non-stationary conditions. Physiol. Meas., 31, 1271.

57.

Malik

, Bigger

J.T.

, Camm

A.J.

, Kleiger

R.E.

, Malliani

, Moss

A.J.

, and Schwartz

P.J.

(1996). Heart rate variability standards of measurement, physiological interpretation, and clinical use. Eur. Heart J., 17, 354–381.

58.

Kurumbanshi

, and Patil

(2013). Performance analysis of monitored biomedical data using biotrace tool. Int. J Ournal Emerg. Technol. Comput. Appl. Sci., 13, 147–153.

59.

Walla

, Brenner

, and Koller

(2011). Objective measures of emotion related to brand attitude: a new way to quantify emotion-related aspects relevant to marketing. PLoS ONE, 6, e26782.

60.

Hughes

P.A.

(2008). Instrumentation for Heart Rate Variability Biofeedback: The NeXus System. Biofeedback. 36.

61.

Nogueira

P.A.

, Rodrigues

, and Oliveira

(2013). Real-time psychophysiological emotional state estimation in digital gameplay scenarios. In: Engineering Applications of Neural Networks. Palmer-Brown

, Dragonova

, Pimenidis

, and Mouratidis

(eds). Springer: Germany, pps. 243–252.

62.

Ting

C.-H.

, Mahfouf

, Nassef

, Linkens

D.A.

, Panoutsos

, Nickel

, Roberts

A.C.

, and Hockey

G.R.J.

(2010). Real-time adaptive automation system based on identification of operator functional state in simulated process control operations. IEEE Trans. Syst. Man Cybern. Part Syst. Hum., 40, 251–262.

63.

McNemar

(1947). Note on the sampling error of the difference between correlated proportions or percentages. Psychometrika, 12, 153–157.

64.

McNemar

(1969). Psychological statistics. Wiley: Hoboken, New Jersey.

65.

Gelman

, Carlin

J.B.

, Stern

H.S.

, and Rubin

D.B.

(2003). Bayesian Data Analysis, Second Edition. CRC Press, LLC.: Boca Raton, Florida.

66.

Witten

I.H.

, and Frank

(2005). Data Mining: Practical Machine Learning Tools and Techniques, Second Edition (Morgan Kaufmann Series in Data Management Systems). Morgan Kaufmann Publishers Inc.: San Francisco, California.

67.

Hempstalk

, Frank

, and Witten

I.H.

(2008). One-class classification by combining density and class probability estimation. In: Machine Learning and Knowledge Discovery in Databases. Springer: New York, pps. 505–519.

68.

Kohavi

(1995). A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection. Morgan Kaufmann: San Francisco, California pps. 1137–1143.

69.

Bekinschtein

T.A.

, Manes

F.F.

, Villarreal

, Owen

A.M.

, and Della Maggiore

(2011). Functional imaging reveals movement preparatory activity in the vegetative state. Front. Hum. Neurosci., 5, 5.

70.

Matthews

S.C.

, Paulus

M.P.

, Simmons

A.N.

, Nelesen

R.A.

, and Dimsdale

J.E.

(2004). Functional subdivisions within anterior cingulate cortex and their relationship to autonomic nervous system function. NeuroImage, 22, 1151–1156.

71.

Critchley

H.D.

, Mathias

C.J.

, Josephs

, O'Doherty

, Zanini

, Dewar

B.-K.

, Cipolotti

, Shallice

, and Dolan

R.J.

(2003). Human cingulate cortex and autonomic control: converging neuroimaging and clinical evidence. Brain, 126, 2139–2152.

72.

Thayer

J.F.

, and Sternberg

(2006). Beyond heart rate variability. Ann. N. Y. Acad. Sci., 1088, 361–372.

73.

Malik

and Camm

A.J.

(2004). Heart rate variability: stress and psychiatric conditions. In: Dynamic Electrocardiography. Berntson

G.G.

and Cacioppo

J.T.

(eds).Wiley: Hoboken, NJ, pps. 57–64.

74.

Hagemann

, Waldstein

S.R.

, and Thayer

J.F.

(2003). Central and autonomic nervous system integration in emotion. Brain Cogn., 52, 79–87.

75.

Friedman

B.H.

(2007). An autonomic flexibility–neurovisceral integration model of anxiety and cardiac vagal tone. Biol. Psychol., 74, 185–199.

76.

Voss

, Heitmann

, Schroeder

, Peters

, and Perz

(2012). Short-term heart rate variability—age dependence in healthy subjects. Physiol. Meas., 33, 1289–1311.

77.

Reardon

, and Malik

(1996). Changes in heart rate variability with age. Pacing Clin. Electrophysiol., 19, 1863–1866.

78.

Pattoneri

, Tirabassi

, Pelà

, Astorri

, Mazzucchi

, and Borghetti

(2005). Circadian blood pressure and heart rate changes in patients in a persistent vegetative state after traumatic brain injury. J. Clin. Hypertens., 7, 734–739.

79.

Bekinschtein

T.A.

, Golombek

D.A.

, Simonetta

S.H.

, Coleman

M.R.

, and Manes

F.F.

(2009). Circadian rhythms in the vegetative state. Brain Inj., 23, 915–919.

80.

Silva

, Alacoque

, Fourcade

, Samii

, Marque

, Woods

, Mazziotta

, Chollet

, and Loubinoux

(2010). Wakefulness and loss of awareness. Neurology, 74, 313–320.

81.

Keren

, Yupatov

, Radai

M.M.

, Elad-Yarum

, Faraggi

, Abboud

, Ring

, and Groswasser

(2005). Heart rate variability (HRV) of patients with traumatic brain injury (TBI) during the post-insult sub-acute period. Brain Inj., 19, 605–611.

82.

Flotta

, Riganello

, and Sannita

W.G.

(2013). Intelligent monitoring of subjects with severe disorder of consciousness. Presented at the Fourth International Conference on Sensor Device Technologies and Applications, Barcelona, Spain.

83.

King

D.R.

, Ogilvie

M.P.

, Pereira

B.M.T.

, Chang

, Manning

R.J.

, Conner

J.A.

, Schulman

C.I.

, McKenney

M.G.

, and Proctor

K.G.

(2009). Heart rate variability as a triage tool in patients with trauma during prehospital helicopter transport: J. Trauma Inj. Infect. Crit. Care, 67, 436–440.

84.

Norris

P.R.

, Morris

J.A.

, Ozdas

, Grogan

E.L.

, and Williams

A.E.

(2005). Heart rate variability predicts trauma patient outcome as early as 12 h: implications for military and civilian triage. J. Surg. Res., 129, 122–128.

85.

Kriegeskorte

, Simmons

W.K.

, Bellgowan

P.S.F.

, and Baker

C.I.

(2009). Circular analysis in systems neuroscience: the dangers of double dipping. Nat. Neurosci., 12, 535–540.

86.

Kung

S.Y.

(2014). Kernel Methods and Machine Learning. Cambridge University Press: Cambridge, UK.

87.

Hand

D.J.

, Mannila

, and Smyth

(2001). Principles of data mining. MIT Press: Cambridge, Massachusetts.