Relationship between frequency spectrum of heart rate variability and autonomic nervous activities during sleep in newborns.

INTRODUCTION: We analyzed the frequency spectrum of two neonatal sleep stages, namely active sleep and quiet sleep, and the relationship between these sleep stages and autonomic nervous activity in 74 newborns and 16 adults as a comparison. METHOD: Active and quiet sleep were differentiated by electroencephalogram (EEG) patterns, eye movements, and respiratory wave patterns; autonomic activity was analyzed using the RR interval of simultaneously recorded electrocardiogram (ECG) signals. Power values (LFa, absolute low frequency; HFa, absolute high frequency), LFa/HFa ratio, and the values of LFn (normalized low frequency) and HFn (normalized high frequency) were obtained. Synchronicity between the power value of HFa and the LFa/HFa ratio during active and quiet sleep was also examined by a new method of chronological demonstration of the power values of HFa and LFa/HFa. RESULTS: We found that LFa, HFa and the LFa/HFa ratio during active sleep were significantly higher than those during quiet sleep in newborns; in adults, on the other hand, the LFa/HFa ratio during rapid eye movement (REM) sleep, considered as active sleep, was significantly higher than that during non-REM sleep, considered as quiet sleep, and HFa values during REM sleep were significantly lower than those during non-REM sleep. LFn during quiet sleep in newborns was significantly lower than that during active sleep. Conversely, HFn during quiet sleep was significantly higher than that during active sleep. Analysis of the four classes of gestational age groups at birth indicated that autonomic nervous activity in a few preterm newborns did not reach the level seen in full-term newborns. Furthermore, the power value of HFa and the LFa/HFa ratio exhibited reverse synchronicity. CONCLUSION: These results indicate that the autonomic patterns in active and quiet sleep of newborns are different from those in REM and non-REM sleep of adults and may be develop to the autonomic patterns in adults, and that parasympathetic activity is dominant during quiet sleep as compared to active sleep from the results of LFn and HFn in newborns. In addition, in some preterm infants, delayed development of the autonomic nervous system can be determined by classifying the autonomic nervous activity pattern of sleep stages.

[An attempt to classify patients with dementia by detrended fluctuation analysis of electroencephalographic data --a preliminary study].

We investigated changes of the scaling exponent alpha estimated by detrended fluctuation analysis (DFA) of electroencephalograms (EEG) in patients with dementia including Alzheimer's disease(AD), and attempted to apply a method of pattern recognition using the alpha value-based feature vector to classify dementia. In 9 patients with AD, 8 patients with other types of dementia (vD), and 7 patients without dementia(C), DFA was performed for approximately one minute with background EEG data recorded at 16 different scalp monopoles. The alpha values were significantly higher in patients with AD at electrodes F7, C3, P3, P4, T3, and T5 than in patients without dementia. No significant difference in alpha values was found between patients with vD and without dementia. Then, an artificial neural network (ANN) was trained on the alpha value-based feature vector of EEG to classify patients with dementia into AD and vD. The trained ANN successfully diagnosed all four new test cases of AD. From these observations, it is suggested that AD has a specific pattern in the alpha value-based feature vector. Thus, pattern recognition using alpha value-based feature vector may be useful for the classification of dementia.

[Relationships among frequency domain and non-linear parameters from heart rate variability].

Frequency domain analysis of heart rate variability (HRV) is used for the evaluation of autonomic activity. Non-linear domain parameters from HRV are also considered useful. However, properties of the latter have not yet been clearly characterized. Therefore, we studied the relationships among the frequency domain and non-linear parameters from HRV. Continuous Holter electrocardiographic monitoring was conducted on 43 healthy female medical staff including laboratory technologists and nurses during an 8-hour working period in our hospital. Low and high frequency components (LF and HF, respectively) of the frequency domain, recurrence rate (REC%) on recurrence plot analysis, scaling exponents al and a2 on detrended fluctuation analysis, and approximate entropy (ApEn) were obtained from HRV. Both the LF/HF ratio and HF were correlated with al and ApEn. REC% was correlated with ApEn and alpha2, whereas alpha2 was correlated only with REC%. Although autonomic parameters from the frequency domain are closely related with some of the non-linear parameters, it is suggested that a2 and REC% reflect different physiological activities.

[Detection of mental task-induced changes in EEG patterns by detrended fluctuation analysis (DFA)].

The detection of mental task-induced changes in electroencephalograms (EEGs) is a challenge. We herein attempted to identify such changes with a long-range correlation parameter, represented as a scaling exponent estimated by detrended fluctuation analysis (DFA). Each of ten volunteers (6 males and 4 females, aged from 23 to 59 years old) was asked to perform two different one minute tasks with an interval between the two while EEGs were recorded: one was to a serial multiplication by 2 (2, 4, 8, 16 ....), and the other, imaginary drawing of a landscape. Five-second segments of EEG data recorded before, during, and after each of the two mental tasks were applied to DFA. The scaling exponent significantly decreased at the right occipital position while imagining the drawing (p = 0.026, by paired t test). Our results suggest that the DFA scaling exponent may be a useful parameter to detect mental task-induced EEG changes.

[Elimination of electrocardiogram artifacts from electroencephalogram records by using the simultaneously recorded electrocardiograph data].

We developed a simple method to eliminate electrocardiogram (ECG) artifacts from electroencephalogram (EEG) records by using simultaneously recorded ECG data. The raw EEG data, the real EEG data and the ECG data were regarded as multi-dimensional vectors Ea, Er and C, respectively. Also, the ECG data, with reduced amplitude whose coefficient was denoted as 'k', were assumed to be overlapped on the real EEG. These assumptions introduced the equations [Ea = Er + k.C], [Er.C = 0] and finally [k = Ea. C/C.C]. This calculation method was implemented by a Macintosh computer using data exported from digital EEG recordings (sampled at 200 Hz with 16-bit resolution). In several subjects, sampling intervals of 5 or 10 seconds for calculation succeeded in eliminating ECG artifacts. However, regardless of the sampling interval, this elimination condition was not always efficient in several other subjects, including a brain-dead patient. It was suggested that the ECG data used were insufficient for the calculation, because only one hand-to-hand reference was used for simultaneous recording, as usual. This one ECG reference was able to express only one ECG projection. Then two other hand-to-foot references of ECG were added to the recordings, and the elimination procedure was performed using all of the simultaneously recorded ECG data at the three references. Consequently, elimination was much improved in most subjects, including the brain-dead patient. Our method may be useful for eliminating ECG artifacts without changing reference electrodes.