Gender differences in adolescent sleep neurophysiology: a high-density sleep EEG study
青少年睡眠神经生理学的性别差异：高密度睡眠脑电图研究

Procedures and EEG data analysis
程序和脑电图数据分析

Sleep EEG was recorded on two consecutive nights (adaptation and baseline) at two time points 6 months apart (Time 1 and Time 2). Forty-nine participants (22 females) completed both sleep EEG assessments (Time 1 and Time 2). The recordings were performed at families’ homes after at least five days of a fixed sleep schedule with 9.5–10 h of time in bed per night ensuring adequate sleep. Compliance to the sleep schedule was measured by means of actigraphy and sleep diaries. Data from the baseline night were used for analyses with the exception of three subjects at Time 1 and three subjects at Time 2 whose baseline night recordings were of insufficient quality and, therefore, data from the adaptation night were used. For two subjects, data from both nights at the first assessment had to be excluded from the analyses due to technical difficulties with the EEG equipment. Therefore, only data from the second assessment time point were available for these subjects.
连续两个晚上（适应和基线）在相隔 6 个月的两个时间点（时间 1 和时间 2）记录睡眠脑电图。 49 名参与者（22 名女性）完成了两项睡眠脑电图评估（时间 1 和时间 2）。这些记录是在至少五天的固定睡眠时间表后在家庭中进行的，每晚在床上的时间为 9.5-10 小时，以确保充足的睡眠。通过体动记录仪和睡眠日记来测量对睡眠时间表的遵守情况。基线夜间的数据用于分析，但时间 1 的三名受试者和时间 2 的三名受试者除外，其基线夜间记录质量不足，因此使用适应夜间的数据。对于两名受试者，由于脑电图设备的技术困难，必须将第一次评估时两个晚上的数据排除在分析之外。因此，这些受试者只能获得第二个评估时间点的数据。

All recordings were performed on a 64-channel Geodesics system (Electrical Geodesic Inc., Eugene, OR, USA). Two channels were used for electrooculogram, two for electromyogram and two for electrocardiogram, resulting in 58 EEG derivations. Data were acquired at 1000 Hz and downsampled to 250 Hz for analysis. Channels with poor quality of signal were excluded based on visual inspection of spectrograms. The signals at all derivations were recalculated to average reference after excluding bad channels. The recordings were scored in 30-s epochs according to standard criteria³⁴. Power density spectra were calculated per epoch (5-s windows; Hanning window; no overlap) using MATLAB (Mathworks, Natick MA, USA). Epochs with artifacts were excluded via a semi-automatic procedure based on the moving average over 21 epochs when power exceeded a threshold in 0.8–4.6 Hz and 20–40 Hz ranges of frequencies as previously described in³⁵. We examined EEG power at each derivation for all frequency bands and the two states—rapid eye movement (REM) and non-rapid eye movement (NREM) sleep. The average number of NREM sleep epochs (30 s per epoch) included in the analyses was 770 (SD = 88) for Time 1 and 743 (SD = 124) for Time 2, while the average number of REM sleep epochs was 277 (SD = 75) for Time 1 and 256 (SD = 62) for Time 2. The following frequency bands were analyzed: delta (1–4.6 Hz), theta (4.8–7.8 Hz), alpha (8–10.8 Hz), sigma (11–16 Hz), beta 1 (16.2–20 Hz), beta 2 (20.2–24 Hz), gamma 1 (24.2–34 Hz) and gamma 2 (34.2–44 Hz). In order to assess gender differences in the topographic distribution of power independent of absolute power differences, we also examined power at each derivation normalized by the total power across derivations.
所有记录均在 64 通道测地线系统（Electrical Geodesic Inc.，尤金，俄勒冈州，美国）上进行。两个通道用于眼电图，两个通道用于肌电图，两个通道用于心电图，产生 58 个 EEG 推导。数据以 1000 Hz 采集，并降采样至 250 Hz 进行分析。根据频谱图的目视检查，排除信号质量差的通道。排除不良通道后，所有衍生信号均重新计算为平均参考值。根据标准³⁴对录音进行 30 秒的评分。使用 MATLAB（Mathworks，Natick MA，美国）计算每个历元的功率密度谱（5 秒窗口；汉宁窗口；无重叠）。当功率超过 0.8–4.6 Hz 和 20–40 Hz 频率范围的阈值时，通过基于 21 个时期的移动平均值的半自动程序排除具有伪影的时期，如先前在³⁵中所述。我们检查了所有频段和两种状态（快速眼动 (REM) 和非快速眼动 (NREM) 睡眠）每次推导时的脑电图功率。分析中包含的 NREM 睡眠时期（每个时期 30 秒）的平均数量在时间 1 中为 770 (SD = 88)，在时间 2 中为 743 (SD = 124)，而 REM 睡眠时期的平均数量为 277 (SD = 124)。 = 75)（时间 1）和 256（SD = 62）（时间 2）。分析了以下频段：delta (1–4.6 Hz)、theta (4.8–7.8 Hz)、alpha (8–10.8 Hz)、sigma ( 11–16 Hz)、beta 1 (16.2–20 Hz)、beta 2 (20.2–24 Hz)、gamma 1 (24.2–34 Hz) 和 gamma 2 (34.2–44 Hz)。 为了评估与绝对功率差异无关的功率地形分布中的性别差异，我们还检查了每个导数的功率，并通过导数的总功率标准化。

In addition to band power, an algorithm based on the envelope of the bandpass filtered signal from the Hilbert transform as described by Rusterholz et al.²⁴ was applied to detect individual spindles in the frequency range between 10 and 16 Hz and extract their main features including amplitude, frequency, duration and density. In addition to the entire spindle range (between 10 and 16 Hz), slow (between 10 and 12 Hz) and fast (between 12 and 16 Hz) spindles were analyzed separately, as differences in the topographic distribution of these two classes of spindles have been shown in adults³⁶ suggesting different mechanisms of generation. In contrast to sigma power which we define as 11–16 Hz, we set the lower limit for the detection of individual spindle events at 10 Hz, because this allows us to detect spindles that are at the lower boundary.
除了频带功率之外，还有一种基于 Hilbert 变换带通滤波信号包络的算法，如 Rusterholz 等人所述。 ²⁴用于检测 10 至 16 Hz 频率范围内的各个纺锤体，并提取其主要特征，包括振幅、频率、持续时间和密度。除了整个主轴范围（10 至 16 Hz 之间）之外，还分别分析了慢速（10 至 12 Hz 之间）和快速（12 至 16 Hz 之间）主轴，因为这两类主轴的地形分布差异在成人中的研究表明³⁶表明了不同的生成机制。与我们定义为 11-16 Hz 的西格玛功率相反，我们将单个纺锤体事件的检测下限设置为 10 Hz，因为这使我们能够检测位于下边界的纺锤体。

Coherence was calculated between all possible channel pairs (i.e., 1653 connections) as |Pxy,(f)|2Pxx(f)Pyy(f) where Pxy(f) is the cross-spectral density and Pxx(f) and Pyy(f) are the auto-spectral density functions of the two signals³⁷. Although a multitude of measures for connectivity have been developed³⁸, we apply coherence as the most established and widely-used in EEG studies. In order to account for contamination through volume conduction, we included only electrode pairs with separations (i.e., arc length) between 10 and 20 cm in our analyses³⁹ resulting in 559 connections.
所有可能的通道对（即 1653 个连接）之间的一致性计算如下： |Pxy,(f)|2Pxx(f)Pyy(f) 在哪里 Pxy(f) 是交叉谱密度， Pxx(f) 和 Pyy(f) 是两个信号³⁷的自谱密度函数。尽管已经开发了多种连通性测量方法³⁸ ，但我们将一致性作为脑电图研究中最成熟和最广泛使用的方法。为了考虑通过体积传导造成的污染，我们仅包括具有分离的电极对（即在我们的分析中，弧长）在 10 到 20 厘米之间³⁹ ，产生 559 个连接。

Age effects 年龄影响

As expected, we found an age-related decline in total sleep time, slow wave sleep as well as a slight, but significant, increase in REM sleep and REM latency (Table ​(Table1).1). Furthermore, similar to previous studies^1–5, we found an age-dependent decline in absolute EEG power across frequencies and sleep states (Figs. 1, ​,2;2; fourth column). Significant age effects on normalized NREM and REM sleep EEG power were found in the delta band, which demonstrated a topographic shift towards frontal regions with increasing age, while alpha, sigma, gamma 1 and gamma 2 (only in NREM sleep) bands showed an age-dependent shift towards posterior regions (Supplementary Figures ¹ and ²). Shorter slow sleep spindle duration, lower spindle amplitude and higher spindle frequency were associated with older age (Fig. 3; fourth column). Coherence was significantly affected by age across bands and the two states with some connections showing a decrease but an overall increase in coherence with age (Fig. 4; second column).
正如预期的那样，我们发现总睡眠时间、慢波睡眠与年龄相关的下降，以及快速眼动睡眠和快速眼动潜伏期略有但显着的增加（表​（表1 ）。1 ）。此外，与之前的研究^{1 – 5}类似，我们发现不同频率和睡眠状态下的绝对脑电图功率随年龄而下降（图1 , ​,2; 2 ; 第四列）。在 delta 带中发现了年龄对标准化 NREM 和 REM 睡眠脑电图功率的显着影响，这表明随着年龄的增长，地形向额叶区域移动，而 alpha、sigma、gamma 1 和 gamma 2（仅在 NREM 睡眠中）带显示了年龄依赖于后部区域的转变（补充图¹和² ）。较短的慢睡眠纺锤波持续时间、较低的纺锤波振幅和较高的纺锤波频率与年龄较大有关（图3 ；第四列）。跨频带的连贯性受到年龄的显着影响，并且具有一些连接的两个状态显示连贯性随着年龄的增长而减少，但总体上增加（图4 ；第二列）。

Pubertal effects 青春期影响

We found no significant effects of pubertal status on absolute (Figs. 1, ​,2;2; fifth column) and minimal effect on normalized (Supplementary Figures ¹ and ²) sleep EEG power for any band or state. However, sleep spindle duration was affected by pubertal status with more advanced pubertal status being associated with shorter spindle duration (Fig. 3; fifth column). Pubertal status had a significant effect on NREM delta, alpha and sigma coherence with a number of connections over central and temporal regions demonstrating a decrease, but the majority of connections showing an increase with advancing pubertal status. We found no effect of pubertal status for the other frequency bands, REM sleep, or any other spindle features.
我们发现青春期状态对任何频段或状态的绝对睡眠脑电图功率没有显着影响（图1、2 ； 2 ；第五列），对标准化（补充图¹和² ）睡眠脑电图功率影响最小。然而，睡眠纺锤波持续时间受到青春期状态的影响，青春期状态越高，纺锤波持续时间越短（图3 ；第五列）。青春期状态对 NREM delta、alpha 和 sigma 一致性有显着影响，中央和颞区的许多连接显示出减少，但大多数连接显示出随着青春期状态的推进而增加。我们发现青春期状态对其他频段、快速眼动睡眠或任何其他纺锤波特征没有影响。

Relatedness effects 相关性效应

Relatedness was significant for sleep latency (F = 8.48; p = 0.005), but no significant effects on any other parameter were observed.
睡眠潜伏期的相关性显着（F = 8.48； p = 0.005），但未观察到对任何其他参数的显着影响。

Sleep EEG power 睡眠脑电图功率

We found a strong effect of gender for absolute sigma power and sleep spindles during NREM sleep, with girls exhibiting greater spindle amplitude, frequency and density across brain regions. Sleep spindles are oscillations in the sigma frequency range during NREM sleep, generated and synchronized by thalamocortical loops. Therefore, they reflect the function of this brain network. The greater NREM sigma activity observed in girls as compared to boys suggests gender differences in thalamocortical circuits. This observation is in line with sleep EEG studies in healthy adults which have shown greater spindle numbers and more sigma power in women as compared to men^15,16,49,50. While we observe greater spindle amplitude, frequency and density in girls, one study of adolescents¹⁸ spanning the ages 12–21 years only found greater fast spindle density in girls and no difference in other spindle features. The discrepancy in findings may be due to the narrow age range in the current study which may provide more sensitivity to detect gender differences in sleep spindle activity, otherwise masked by the overall developmental pattern across a broader age range. Furthermore, the previous study assessed spindle activity at one frontal (F3; slow spindles) and one central (C3; fast spindles) derivation which limits their ability to detect local gender differences. Although previous studies have shown topographic differences between slow and fast spindles³⁶ suggesting different generators, the gender differences we observed were consistent between the two spindle classes and widespread in topography.
我们发现性别对 NREM 睡眠期间的绝对西格玛功率和睡眠纺锤波有很强的影响，女孩在整个大脑区域表现出更大的纺锤波振幅、频率和密度。睡眠纺锤波是 NREM 睡眠期间西格玛频率范围内的振荡，由丘脑皮质环产生和同步。因此，它们反映了这个大脑网络的功能。与男孩相比，女孩的 NREM 西格玛活动更大，这表明丘脑皮质回路存在性别差异。这一观察结果与健康成年人的睡眠脑电图研究一致，该研究表明，与^{15、16、49、50岁的男性}相比，女性的纺锤体数量和西格玛功率更高。虽然我们观察到女孩的纺锤波振幅、频率和密度较高，但一项针对¹²至 21 岁青少年的研究仅发现女孩的快速纺锤波密度较高，而其他纺锤波特征没有差异。研究结果的差异可能是由于当前研究的年龄范围较窄，这可能为检测睡眠纺锤波活动的性别差异提供了更高的敏感性，否则会被更广泛年龄范围的整体发育模式所掩盖。此外，之前的研究评估了一个额叶（F3；慢纺锤体）和一个中枢（C3；快纺锤体）衍生的纺锤体活动，这限制了他们检测局部性别差异的能力。 尽管之前的研究显示慢速和快速纺锤体之间的地形差异³⁶表明不同的发生器，但我们观察到的性别差异在两个纺锤体类别之间是一致的，并且在地形上广泛存在。

The observed differences in sleep spindle activity between boys and girls may be of functional significance. For example, the well-established association between sleep spindle activity and memory consolidation/intelligence has been shown to differ between genders^50–52. In an adolescent sample, Bódizs et al.⁵¹ found positive associations between spindle density as well as amplitude and fluid IQ in adult females, while in adult males greater fluid IQ was associated with higher spindle frequency. The authors suggest different cognitive strategies between males and females as a possible explanation.
观察到的男孩和女孩之间睡眠纺锤波活动的差异可能具有功能意义。例如，睡眠纺锤波活动与记忆巩固/智力之间已确立的关联已被证明在性别之间存在差异^{50 – 52} 。在青少年样本中，Bódizs 等人。 ⁵¹发现成年女性的纺锤体密度和振幅与液体智商之间呈正相关，而在成年男性中，较高的液体智商与较高的纺锤体频率相关。作者提出男性和女性之间不同的认知策略作为可能的解释。

Another possible functional implication of our findings relates to the role of sleep spindles in sleep protection^53,54. Several empirical studies have shown that these oscillations can predict the ability to maintain sleep^55–57. Dang-Vu et al.⁵⁶ found that subjects demonstrating greater sleep spindle activity were more tolerant towards external sounds during sleep, i.e. they were able to stay asleep at higher levels of noise. Extrapolating from this, more spindle activity would suggest more protected sleep in females as compared to males. In support of this, we observed more wake after sleep onset in boys suggesting gender differences in sleep protection mechanisms. We note that this finding somewhat contradicts the current literature which reports higher rates of insomnia and more sleep difficulties in adult women. In fact, we find that subjective sleep quality as measured by the Sleep Habits Survey⁵⁸ did not differ between boys and girls in our sample, and previous studies have suggested that gender differences in subjective sleep quality do not emerge till the end of puberty⁵⁹. Finally, indirect evidence for gender differences in the thalamocortical system comes from waking studies which report gender differences in thalamocortical processing of auditory gating⁶⁰ and pain perception⁶¹.
我们的研究结果的另一个可能的功能意义与睡眠纺锤体在睡眠保护中的作用有关^{53、54 。}多项实证研究表明，这些振荡可以预测维持睡眠的能力^{55 – 57} 。 Dang-Vu 等人。 ⁵⁶发现，睡眠纺锤波活动较大的受试者在睡眠期间对外部声音的耐受性更强，即他们能够在较高水平的噪音下保持睡眠状态。由此推断，与男性相比，更多的纺锤体活动表明女性的睡眠受到更多保护。为了支持这一点，我们观察到男孩入睡后醒来的次数更多，这表明睡眠保护机制存在性别差异。我们注意到，这一发现在某种程度上与当前文献报道的成年女性失眠率较高和睡眠困难较多有关相矛盾。事实上，我们发现，通过睡眠习惯调查测量的主观睡眠质量⁵⁸在我们样本中的男孩和女孩之间没有差异，并且之前的研究表明，主观睡眠质量的性别差异直到青春期结束时才会出现⁵⁹ 。最后，丘脑皮质系统性别差异的间接证据来自清醒时的研究，这些研究报告了丘脑皮质听觉门控⁶⁰和疼痛感知⁶¹处理过程中的性别差异。

In addition to NREM sleep sigma power, girls in our sample exhibited greater absolute sleep EEG power in higher frequencies (beta 1 to gamma 2) for both NREM and REM sleep in line with several resting-state EEG studies in adults^62–64. Because we find gender differences in high frequencies for both NREM and REM sleep, two states which are functionally and biologically unique⁶⁵, we hypothesize that anatomical differences underlie the greater high-frequency power in females. As it is still unclear what brain structures are associated with beta and gamma power, future work should address this issue.
除了 NREM 睡眠西格玛功率外，我们样本中的女孩在 NREM 和 REM 睡眠的较高频率（β 1 至 γ 2）中表现出更大的绝对睡眠脑电图功率，这与几项成人静息态脑电图研究一致^{62 – 64} 。因为我们发现 NREM 和 REM 睡眠的高频存在性别差异，这两种状态在功能和生物学上都是独特的⁶⁵ ，所以我们假设解剖学差异是女性更高频率功率的基础。由于目前尚不清楚哪些大脑结构与贝塔和伽马能量相关，未来的工作应该解决这个问题。

In contrast, the lower frequency bands (< 11 Hz) did not show differences in absolute sleep EEG power between boys and girls. The delta band has previously been shown to correlate with gray matter volume in the anterior cingulate and orbital frontal cortex⁶⁶ and mirror the reduction of cortical gray matter volume in this age range^12,13. As the age range in our sample is rather narrow and selected to reflect the period of most rapid changes in sleep EEG power⁶, gender differences in this frequency range may have been masked as boys start the pubertal decline of power later than girls. Therefore, our data set may constitute the last ascending part of the developmental trajectory in boys, while delta/theta power in girls is on the descending part of the maturational curve. This idea is supported by the interaction between the effects of gender and age we observed in the lower frequencies with power declining at older ages in boys as compared to girls. Furthermore, the large degree of inter-individual variability in the trajectories of both brain development⁶⁷ and sleep EEG power⁶⁸ further complicates this issue as some individuals may reach peak neuronal density/EEG power earlier than others. Examining developmental trajectories might, therefore, provide new insights into the nature of gender differences and should be taken into account in future studies.
相比之下，较低频段（< 11 Hz）男孩和女孩之间的绝对睡眠脑电图功率没有差异。先前已证明 Delta 带与前扣带皮层和眶额皮质⁶⁶中的灰质体积相关，并反映了该年龄范围^12、13中皮质灰质体积的减少。由于我们样本中的年龄范围相当窄，并选择反映睡眠脑电图功率变化最快的时期⁶ ，因此该频率范围内的性别差异可能被掩盖，因为男孩比女孩更晚开始青春期功率下降。因此，我们的数据集可能构成男孩发育轨迹的最后上升部分，而女孩的 delta/theta 功率位于成熟曲线的下降部分。这一观点得到了我们在较低频率中观察到的性别和年龄影响之间相互作用的支持，与女孩相比，男孩年龄较大时功率下降。此外，大脑发育⁶⁷和睡眠脑电图功率⁶⁸的轨​​迹存在很大程度的个体间差异，这使得这个问题进一步复杂化，因为有些人可能比其他人更早达到峰值神经元密度/脑电图功率。因此，检查发展轨迹可能会为性别差异的本质提供新的见解，并应在未来的研究中予以考虑。

For normalized sleep EEG power, we observed a topographic shift towards frontal regions in girls across frequencies and sleep states. Sleep studies in adults have shown attenuated differences between females and males when power is normalized suggesting that such differences are due to structural effects such as skull thickness⁶⁹. However, our findings in adolescence do not provide support for this hypothesis. In our study, normalized power reveals additional differences related to topographic shifts previously masked by pronounced differences in absolute power. The topographic shift towards frontal regions in girls may be a reflection of advanced brain maturation according to maturational trajectories observed for topographic distribution of sleep EEG activity across childhood and adolescence²².
对于标准化睡眠脑电图功率，我们观察到女孩在不同频率和睡眠状态下，地形向额叶区域发生转变。成人睡眠研究表明，当能量正常化时，女性和男性之间的差异减弱，这表明这种差异是由于头骨厚度等结构效应造成的⁶⁹ 。然而，我们在青春期的发现并没有为这一假设提供支持。在我们的研究中，归一化功率揭示了与先前被绝对功率的明显差异所掩盖的地形变化相关的额外差异。根据对儿童期和青春期睡眠脑电图活动的地形分布观察到的成熟轨迹，女孩向额叶区域的地形转变可能反映了大脑提前成熟²² 。

Sleep EEG coherence 睡眠脑电图一致性

With regards to sleep EEG coherence, a measure of functional connectivity, we observed greater coherence in girls than boys across bands and states with the exception of NREM alpha coherence, where boys demonstrated higher values. These results are in agreement with previous waking EEG coherence studies in children and adolescents^28,29,70. A recent sleep EEG study in adults found similar gender differences in the alpha and sigma bands during NREM sleep²⁷ suggesting that these gender differences prevail over the life course. However, in the beta band the authors found the opposite gender effect with men exhibiting stronger connectivity than women during both NREM and REM sleep. We conjecture that it might be developmental changes that underlie the differences between this study and our results. Larger coherence values in girls have previously been hypothesized to reflect a maturational lead in females⁷⁰. Interestingly, we find the largest differences between males and females in NREM delta and sigma power, two oscillations which rely on corticocortical and thalamocortical connectivity^14,71–74. Both greater anatomical connectivity⁷⁵ as well as greater functional connectivity^76,77 have been observed for women in MRI studies. In line with our findings, a large MRI study of adult subjects found not only higher functional connectivity in women than in men, but also that gender differences were maximal in the anterior thalamus⁷⁷. The authors suggest that greater connectivity in women may facilitate functions relying on synchronization between wide brain areas and emphasize the implication of such findings for psychiatric disorders associated with altered brain connectivity⁷⁷. Taken together, gender differences in sleep EEG parameters vary between frequency bands^78,79 and are therefore more likely to reflect functional differences in brain networks¹⁶ than merely differences in brain structure.
关于睡眠脑电图一致性（功能连通性的衡量标准），我们观察到，在各频段和状态中，女孩的一致性高于男孩，但 NREM α 一致性除外，其中男孩表现出更高的值。这些结果与之前针对儿童和青少年进行的清醒脑电图一致性研究一致^{28 , 29 , 70} 。最近的一项针对成人的睡眠脑电图研究发现，NREM 睡眠期间 α 和 Sigma 频带存在类似的性别差异²⁷ ，这表明这些性别差异在整个生命过程中普遍存在。然而，在 beta 频段，作者发现了相反的性别效应，男性在 NREM 和 REM 睡眠期间表现出比女性更强的连接性。我们推测，这项研究与我们的结果之间的差异可能是发育变化造成的。此前曾假设女孩具有较大的一致性值，以反映女性的成熟领先性⁷⁰ 。有趣的是，我们发现男性和女性之间最大的差异在于 NREM delta 和 sigma 功率，这两种振荡依赖于皮质和丘脑皮质连接^{14 , 71 – 74} 。在 MRI 研究中观察到女性具有更大的解剖连接性⁷⁵和更大的功能连接^性76、77 。 根据我们的研究结果，一项针对成人受试者的大型 MRI 研究发现，女性不仅比男性具有更高的功能连接，而且前丘脑的性别差异最大⁷⁷ 。作者认为，女性更强的连通性可能会促进依赖于广泛大脑区域之间同步的功能，并强调这些发现对与大脑连通性改变相关的精神疾病的影响⁷⁷ 。总而言之，睡眠EEG参数中的性别差异在频带^{78、79之间}变化，并且因此更有可能反映大脑网络¹⁶中的功能差异而不仅仅是大脑结构中的差异。

Impact of age and puberty
年龄和青春期的影响

Age had a significant influence on power and coherence across frequency bands and sleep states, which is likely due to anatomical changes to the brain^8,80. On the other hand, we did not find an effect of pubertal status (with the exception of low-frequency coherence during NREM sleep). This finding is in agreement with results from a longitudinal study by Feinberg et al.²⁰ showing that the association between the rate of the NREM delta power decline and the rate of pubertal development disappears when controlling for age. The authors hypothesize that the power decline is associated with developmental processes other than physical or sexual maturation. Further evidence for this notion comes from a study investigating delta power in girls with central precocious puberty⁸¹ concluding that increased estrogen does not cause the adolescent decline in delta power.
年龄对频段和睡眠状态的功率和一致性有显着影响，这可能是由于大脑的解剖变化^{8 , 80} 。另一方面，我们没有发现青春期状态的影响（NREM 睡眠期间的低频一致性除外）。这一发现与 Feinberg 等人的纵向研究结果一致。^{图 20}表明，当控制年龄时，NREM δ功率下降速率与青春期发育速率之间的关联消失。作者假设，功率下降与身体或性成熟以外的发育过程有关。这一观点的进一步证据来自一项调查中枢性性早熟女孩 Delta 功率的研究^81，该研究得出的结论是雌激素增加不会导致青少年 Delta 功率下降。

We thank Dr. Christoph Hamann, Nathaline Margot, Daniela Rupp, Tammy Timmers and Julia Hegy for data collection, and Dr. Thomas Rusterholz for help with data processing. We would also like to thank Professor Peter Achermann for valuable scientific input, and Professors Alexander Borbély and Irene Tobler for inspiring discussions. We are grateful to the participants and their families for taking part in the study.
我们感谢 Christoph Hamann 博士、Nathaline Margot、Daniela Rupp、Tammy Timmers 和 Julia Hegy 收集数据，感谢 Thomas Rusterholz 博士在数据处理方面提供的帮助。我们还要感谢 Peter Achermann 教授提供的宝贵科学投入，以及 Alexander Borbély 和 Irene Tobler 教授启发性的讨论。我们感谢参与者及其家人参与这项研究。