Insomnia Treatment Assessment Based on Physiological Data Analysis

Share Embed


Descrição do Produto

Proceedings of the 29th Annual International Conference of the IEEE EMBS Cité Internationale, Lyon, France August 23-26, 2007.

SuB10.3

Insomnia Treatment Assessment Based on Physiological Data Analysis Ioanna Chouvarda, Member, IEEE , Christos Papadelis, Nathalie Domis, Luc Staner, and Nicos Maglaveras, Senior Member, IEEE

Abstract—For the purposes of insomnia treatment, pharmacotherapy is widely used, despite the possibility for the use of behavioural treatment of insomnia. Thus, the assessment and treatment of patients with insomnia needs further investigation. This work addresses insomnia treatment evaluation and medication side-effect assessment based on continuous physiological signals such as EEG and ECG monitoring and analysis. EEG and ECG measurements regarding drug medication (verum/placebo cases) have been used in a series of experiments, where spectral and non-linear features have been calculated, for assessing a possible distinct behaviour between the verum/placebo condition and furthermore the relation of features to a physiological conditions. Results show that a combination of EEG and ECG based characteristics, both spectral and non-linear, can be used to reveal the differences introduced with insomnia medication treatment, either being improvement in the hyperarousal state, or undesired side effects.

I

I. INTRODUCTION

nsomnia is often defined as sleep onset or maintenance problem. Although increased wakefulness is a key factor, detailed patterns of wakefulness in insomnia patients, either during the day or night have not been extensively studied. Apart from patient interview and physical examination performed by the physician, polysomnography is the diagnostic tool most often employed to collect further information about insomnia, and specifically about the sleep patterns. As far as insomnia treatment is concerned, pharmacotherapy remains the most widely used treatment, despite the documented efficacy of behavioural treatments for insomnia. However, there is often a discrepancy between subjective reports of daytime functioning deficits and objective impairments among individuals with insomnia complaints, in both the medicated and unmedicated cases. The assessment and treatment of patients with insomnia needs therefore a closer look [1]. Thus, the work presented focuses on insomnia treatment evaluation and medication side-effect assessment based on continuous physiological signals monitoring and analysis. Manuscript received April 2, 2007. This work was supported in part by the EU-IST - 2003- 507231 project SENSATION. I. Chouvarda, C. Papadelis and N. Maglaveras are with .Lab of Medical Informatics, The Medical School, Aristotle University of Thessaloniki, 54124 Thessaoniki, Greece ( e-mail: [email protected], [email protected] and [email protected]). N. Domis, and L. Staner are with FORENAP, 68250 Rouffach, France ( e-mail : {nathalie.domis, luc.staner}@forenap.com }

1-4244-0788-5/07/$20.00 ©2007 IEEE

EEG and ECG measurements regarding drug medication (verum/placebo cases) are used in a series of experiments, where spectral and non-linear features have been calculated, for assessing a possible distinct behaviour between the verum/placebo condition and furthermore the relation of features to a physiological condition. II. METHODS EEG and ECG data have been recorded after drug intake in verum and placebo conditions, covering the whole time 24 hour period of the insomnia medication treatment, until next morning. The way various signals characteristics differentiate is examined during the whole daytime. A. EEG analysis While EEG data have been traditionally used for sleep staging, the entire 24-hour EEG activity is of interest here. Spectral analysis as well as non-linear analysis has been applied, for the assessment of the differences between insomniacs and normal population during different times of the day, or for drug assessment. Specifically, the features under consideration were the following: --Power Spectrum band features, i.e. Power Spectrum and average in the frequency bands (normalised by the total). EEG bands taken into consideration for power spectrum analysis were: band1=1-3.5Hz (delta – deep sleep), band2=3.5-7.5Hz (theta –drowsiness), band3=7.510.5Hz (alpha1-relaxation), band4=10.5-12.5Hz (alpha2concentration), band5=12.5-17.5Hz (beta1-alertness, focus) and band6=17.5-22Hz (beta2-alertness, agitation). --ASI - alpha slow-wave index. ASI=alpha/(delta + theta), related to the arousal level [2]. --Coherence between relevant left and right channels. The maximum value per band was considered as the feature in use. Coherence or cross-spectral analysis, may be used to identify variations which have similar spectral properties (high power in the same spectral frequency bands), i.e. if the variability of two distinct time series is interrelated in the spectral domain. --Transfer function of adjacent left and central channels. The relationship between the input channel x and output channel y is modeled by the linear, time-invariant transfer function Txy, revealing possible functional connectivity patterns. The maximum value per band was considered as the feature in use. --Sample entropy in each channel. Larger sample

6694

Authorized licensed use limited to: IEEE Xplore. Downloaded on November 5, 2008 at 11:43 from IEEE Xplore. Restrictions apply.

entropy values indicate greater independence, less predictability, hence greater complexity in the data [3].

Fig. 1. Cumulative number of placebo/verum statistical differences among electrodes by frequency bands for different times of the day.

B. ECG Analysis ECG, Heart rate variability (HRV), blood pressure variability (BPV), and the assessment of baroreflex sensitivity are widely accepted methods for analyzing and characterizing cardiovascular regulation and for an enhanced risk evaluation in different diseases. Changes in autonomic nervous system (ANS) tone are known to accompany various mental disorders [4]. The ideas behind this analysis concern the assessment of drug insomnia treatment, and specifically to check how effective each drug is. The side-effects and the drawbacks in the subject’s quality of life, due to insomnia medication, consist another important aspect. This biosignals-based assessment can take place in parallel with subject’s subjective reporting, consisting the objective markers of the treatment efficiency. After robustly extracting the heart rate from the ECG time-series, a number of different features have been calculated in each segment, including: --Time-based features: Mean and standard deviation of the heart rate --Linear Spectral analysis of the Heart Rate: Power Spectrum and energy in LF, HF, VLF band, as well as the band energy ratios, where VLF=0.003-0.04 Hz, LF=0.040.15Hz and HF=0.15-0.4Hz. --Non-linear Features of the RR intervals: Sample entropy of the RR time-series, corresponding to signal complexity, and detrended fluctuation analysis of the RR time-series, useful in revealing the extent of long-range correlations in time series [3], [5]. III. RESULTS A. EEG Analysis for Treatment Assessment The power spectral analysis revealed important differences between the verum and placebo group, along the whole daytime, as shown by statistical paired tests applied for the signal features acquired at the same time. Specifically, during the day, an increase in delta band,

decrease in alpha band, and increase in beta band were observed. Band energy differences, as quantified by the number of channels/bands were statistically significant differences are found, increase in the morning, decrease afternoon, in order to increase again at night, while differences are minor in the next morning (Fig 1). In the work of Merica et al. [6] regarding spectral characteristics of the sleep EEG of insomniacs and healthy subjects, beta power is reported as significantly increased in NREM unlike the lower bands, while in REM, insomniacs show lower levels in the delta and theta bands, whereas power in the faster frequency bands is significantly increased. ASI feature indicating arousal has some decrease with verum, as compared to the placebo case, which may express the decrease of hyperarousal related to insomnia, after medication treatment. Sample entropy (complexity) is decreased in the morning for the verum as compared to the placebo, while in afternoon/night it increases with verum. The differences are not statistically significant. The verum/placebo difference is minor after 24 hours (next morning). Coherence and transfer functions were calculated in order to see possible spatial functioning changes between placebo and verum, at different times of the day. Altered interhemispheric synchronization patterns have been reported in relation to various pathologic conditions, while previous research has demonstrated an association between EEG coherence and sleep [7]. It has been reported that interhemispheric and intrahemispheric coherence in normal subjects are greater during all sleep stages when compared to wakefulness, and it depends on the time of the day. Here, Coherence is calculated in left-right corresponding channels, ie Fp1-Fp2…C3-C4,.. P3-P4…O1-O2 and for most frequency bands, antero-posterior differences were found in left/right hemisphere coherence. In the front channels (FP1 FP2 , F7 -F8, F3 - F4 ) there is a decreased coherence with verum, while in other channels (T3-T4, C3-C4, T5-T6, P3P4), an increased left-right coherence is observed with verum, however coherence level is small in both placebo/verum cases. Decrease in Fp1-Fp2 and increase in T3-T4 coherence is also evident next morning. Adjacent channel transfer function shows changes during the day, mostly decrease in the alpha band. This decreased connectivity pattern may be related to the altered arousal state and drowsiness. Fz-Cz and Pz-Oz connectivities are always decreased with verum, in the alpha-beta bands. In all cases, except delta-theta next morning, Cz-Pz and C3-P3 have an increased connectivity with verum. Most of the statistically significant connectivity differences occur in the morning or at least before 15.00. During the day, the alpha rhythm differences show decreased connectivity. Next morning there are almost no differences. Anterio-posterior differences in the functional connectivity between pre-sleep and sleep phases have been reported by [8], who also observed differences between the frontal-occipital flow patterns and the frontal-parietal flow patterns in the

6695 Authorized licensed use limited to: IEEE Xplore. Downloaded on November 5, 2008 at 11:43 from IEEE Xplore. Restrictions apply.

presleep-sleep phases. B. ECG Analysis for Drug Assessment The Power spectrum HR analysis leads to interesting observations. In the morning upon drug intake, there is an increase in VLF components, also observed at night, but this effect is eliminated next morning.. Heart Rate VHF components are believed to be related with parasympathetic and circadian rhythms. On the other hand, there is no statistical difference for LF components (related to the sympathetic activity), although there is a decrease at night with verum. Regarding the HF components and vagal modulation, there is an initial decrease in HF with verum, and also a decrease during the night, but not next morning. Furthermore, in the morning there is difference in the VLF/LF and in the LF/HF ratios, both being higher with verum. Next morning this difference is not observed.

and increases both dfa features [9]. Decrease in fixed respiration rate, increases dfa fast and decreases dfa slow and entropy, whereas changes in dfa are also related to the sleep stages [10]. Dfa changes are reported in pathologic conditions, such as sleep apnea, heart failure, etc [11]. Overall, based on the observations drawn by the HR analysis, the drug has a clear effect when taken, which is not obvious next morning, changing the sleep related patterns. IV. CONCLUSION The analysis described in this work shows that a combination of EEG and ECG characteristics, both spectral and non-linear, can reveal the differences introduced with insomnia medication treatment, either being improvement in the hyperarousal state, or unwanted side-effects. Among the future works for the completion of this study is the development of a complete decision support system. REFERENCES [1]

Fig. 2. Evolution with time of relative difference in the mean HR DFA fast and mean DFA slow between placebo and verum

Regarding the time-domain features of Heart rate, the average HR shows statistical difference (is higher with verum) in the morning when drug is delivered. At night these changes, maybe denoting a deeper sleep, and in the morning HR is again higher, but significantly. For HR standard deviation, there is statistical difference in the morning, but also after bedtime, and early in the next morning. With verum, HR standard deviation tends to be higher. Dfa features show differences between placebo and verum, even early next morning (Fig 2). Statistically significant time periods are in the morning (11-12), at late night and early next morning. Specifically, Dfa slow falls with verum at night. Dfa fast is generally higher, and specifically in the morning and after bedtime. The decrease in dfa-slow, implies a change in system dynamics, towards more irregular forms in the long-range correlations, while the increase in the dfa-fast feature shows smoother short term dynamics. A tendency for decrease in sample entropy, ie towards more regular patterns of HR, is observed with verum, especially in the morning and at bedtime (23.001.00). In the literature, vagal blockade decreases entropy

A. Vignola, C. Lamoureux, C.H. Bastien, and C.M. Morin, “Effects of chronic insomnia and use of benzodiazepines on daytime performance in older adults,” J Gerontol B Psychol Sci Soc Sci., vol 55, no 1, Jan 2000 , pp:54-62. [2] M. Jobert M, H. Schulz , P. Jahnig, C. Tismer, F. Bes, and H. Escola H., “A computerized method for detecting episodes of wakefulness during sleep based on the alpha slow-wave index (ASI),” Sleep, vol 17, no 1.Feb 1994, pp::37-46. [3] J.S. Richman and J.R. Moorman, “Physiological time-series analysis using approximate entropy and sample entropy,”. Am J Physiol Heart Circ Physiol, vol. 278, issue 6, 2000, pp:H2039-H2049. [4] A. Birkhofer , G. Schmidt, and H. Forstl, “Heart and brain -- the influence of psychiatric disorders and their therapy on the heart rate variability”, Fortschr Neurol Psychiatr., vol 73, no 4, Apr 2005 pp:192-205. [5] C-K Peng, J.M. Hausdorff, and A.L. Goldberger , “Fractal mechanisms in neural control: Human heartbeat and gait dynamics in health and disease”. In: Walleczek J, ed. Nonlinear Dynamics, SelfOrganization, and Biomedicine. Cambridge: Cambridge University Press, 1999. [6] H. Merica, R. Blois, and J.M. Gaillard, “Spectral characteristics of sleep EEG in chronic insomnia,” European Journal of Neuroscience, vol.10 , no 5, 1998, pp: 1826–1834. [7] H. Von Gizycki, G. Jean-Louis, F Zizi, C Stampi, H Green, S Franconeri, AJ Spielman , J Nunes, and Taub, “Time of day effect on alpha band interhemispheric coherence,” Sleep Research, 1996, pp:25: 39. [8] L De Gennaro, F Vecchio, M Ferrara, G Curcio, PM Rossini, and C Babiloni. “Changes in fronto-posterior functional coupling at sleep onset in humans”, J Sleep Res, Sep. 2004, vol 13, no 3, pp: 209-17. [9] J. Penttila,, A. Helminen, T. Jartti, T. Kuusela, H.V. Huikuri, M.P. Tulppo, and H. Scheinin, “Effect of cardiac vagal outflow on complexity and fractal correlation properties of heart rate dynamic,” Autonomic & Autacoid Pharmacology, vol 23, no 3, June 2003, pp:173-179. [10] R. Krishnam, S. Chatlapalli, H. Nazeran, E. Haltiwanger, and Y. Pamula, “Detrended Fluctuation Analysis: A Suitable Long-term Measure of HRV Signals in Children with Sleep Disordered Breathing”, 27th Annual International Conference of the IEEEEMBS, 01-04 Sept. 2005, Page(s): 1174 – 1177. [11] T. Penzel, J.W. Kantelhardt, H.F. Becker, J.H. Peter, and A.Bunde, “Detrended fluctuation analysis and spectral analysis of heart rate variability for sleep stage and sleep apnea identification” Computers in Cardiology, 21-24 Sept. 2003, Page(s): 307 – 310.

6696 Authorized licensed use limited to: IEEE Xplore. Downloaded on November 5, 2008 at 11:43 from IEEE Xplore. Restrictions apply.

Lihat lebih banyak...

Comentários

Copyright © 2017 DADOSPDF Inc.