Jiangxi Puzhuo Medical Equipment Co., Ltd

The teenage years represent a golden stage for brain development and cognitive shaping. Sleep, as a crucial foundation of life activities, has always been a focal point of research in the fields of neuroscience and pediatrics due to its intrinsic correlation with brain structure and functional development.

Professor Cheng Wei's team from Huashan Hospital, Fudan University, in collaboration with multiple universities both domestically and internationally, published a landmark study in Cell Reports. This study relied on objective sleep monitoring using wearable devices and multimodal brain imaging technology to conduct a systematic and longitudinal study on 3,222 adolescents aged 9-14. It clearly revealed two core dimensions of correlation between sleep characteristics and brain development, identified three distinct sleep-brain biotypes with significant differences, and confirmed that the differences in brain and cognitive development among these biotypes persist in early adolescence.

This study not only provides a precise theoretical basis for sleep intervention in adolescents, but also highlights the core value of wearable medical devices in sleep health monitoring and clinical research, pointing out a new direction for the development and application of sleep-related medical devices.

Research Design

The core data for this study are all sourced from the internationally renowned Adolescent Brain Cognitive Development (ABCD) study. The study design combines large sample size, objectivity, and rigor: using 3,222 adolescents aged 11-12 as the discovery sample, and completing dual replication verification through the full four-year follow-up data of 1,190 adolescents aged 13-14 and 596 independent follow-up data without overlap.

The study employed the Fitbit wearable wristband to objectively record the sleep of participants for three consecutive weeks, collecting a total of 18 core sleep indicators. Additionally, it integrated 85 brain volume indicators (68 cortical regions + 17 subcortical regions) and 282 brain functional connectivity indicators (including intra-cortical, inter-cortical, and cortico-subcortical connections). By utilizing advanced machine learning methods such as sparse canonical correlation analysis (sCCA) and hierarchical clustering, the study achieved high-dimensional and multi-dimensional correlation analysis between sleep characteristics and brain structure and function. Both the comprehensiveness of the research data and the progressiveness of the analytical methods are at the forefront of the field.

Figure 1:Overview of research design and analysis This study utilizes objective sleep characteristics based on devices and brain imaging measurement methods (covering brain structure and function) to explore the neural correlates of adolescent sleep and its psychophysiological significance. All data were sourced from the Adolescent Brain Cognitive Development (ABCD) study. The findings of the sample were based on 2-year follow-up data from ages 11-12 (n=3,222) and were validated twice using 4-year follow-up data (n=1,190; ages 13-14).

Core Result 1: Two high-dimensional sleep-brain correlation dimensions accurately depict the differentiated regulatory patterns of sleep on the brain

Through sCCA analysis, the study successfully identified two potentially significant sleep-brain correlation dimensions with statistical significance in high-dimensional data, clarifying the specific effects of different combinations of sleep characteristics on brain structure and function. This fills the gap in traditional single-dimensional and single-indicator research and provides a new perspective for understanding the neural mechanisms of sleep regulating brain development.

Figure 2: The correlation between sleep brain region dimensions and sleep patterns jointly influences the characteristics of related brain regions.

(A) Sparse Canonical Correlation Analysis (sCCA) reveals the multivariate patterns of two correlation dimensions between sleep and brain regions. The scatterplot illustrates the canonical variables of brain regions and sleep (which are linear combinations of brain imaging indicators, including brain volume and functional network connectivity),andThe correlation between objective sleep characteristics. In the discovered sample, the canonical correlation of all two dimensions was significantly higher than the empirical null distribution, with p-values < 0.05 after adjustment for age and gender.

(B) Define two sleep dimensions: the left side represents the "fall asleep time-total duration" dominant mode, and the right side represents the "heart rate-light sleep duration" dominant mode. Each sleep index on the y-axis is arranged in descending order according to the typical sleep load values shown on the x-axis. The statistically significant loads determined through the resampling procedure are marked in orange (positive weight) and green (negative weight), and gray circles indicate non-significance.

Dimension 1: Sleep latency - total duration as the dominant pattern, with a core impact on cortical-subcortical functional connectivity

This dimension is characterized by core sleep features such as later bedtime, reduced total sleep duration, and increased frequency of nocturnal awakenings. Closely associated with these features is a widespread decrease in subcortical-cortical functional connectivity, accompanied by a compensatory increase in a small amount of functional connectivity within and between primary cortical networks.

From the perspective of affected brain regions and functional networks, the core involves the sensorimotor hand network, the cingulate-insular network, as well as key subcortical regions such as the basal ganglia, amygdala, hippocampus, and thalamus. Among them, the functional connectivity between the sensorimotor hand network and the globus pallidus decreases most significantly, while the functional connectivity between the sensorimotor hand network and the sensorimotor mouth network increases most notably.

This result clearly indicates that long-term late sleep and insufficient sleep do not simply affect the "rest" of the brain, but directly weaken the signal transmission and coordination between the cortical and subcortical core brain regions. The abnormal connectivity of the primary sensorimotor network may further affect the basic neural activity, body coordination, and information processing ability of adolescents. This also provides a neurological explanation for the phenomena of inattention and slow response caused by insufficient sleep during adolescence.

Dimension 2: Heart rate-light sleep duration dominant mode, the first confirmation of the direct effect of sleep heart rate on brain volume

This is an important new finding of this study. This dimension is characterized by high heart rate at all sleep stages and significantly shortened light sleep duration. It is associated with both reduced brain volume and decreased cortico-subcortical functional connectivity. This study is the first in academia to confirm the direct correlation between sleep heart rate indicators and adolescent brain structural development, and improves the synergistic regulatory mechanism of "heart-brain-sleep".

At the brain volume level, the volumes of the hippocampus, lateral occipital lobe, lingual gyrus, supramarginal gyrus, amygdala and other regions are significantly reduced. These brain regions are closely related to memory storage, visual processing and emotion regulation, and the reduction in hippocampal volume is directly associated with the development of learning and memory ability. At the functional connectivity level, the connections between core functional networks such as the sensorimotor hand/mouth network, cingulate-parietal network and salience network, and subcortical regions including ventral diencephalon, basal ganglia and brainstem are significantly weakened, among which the reduction in connectivity between the sensorimotor mouth network and ventral diencephalon is the most prominent.

The research team pointed out that high heart rate during sleep not only reflects abnormalities in autonomic nervous function, but also directly affects the normal development of brain structure and cross-regional functional coordination. As an important part of the sleep cycle, insufficient light sleep duration will further aggravate this negative impact. This also provides a potential basis for evaluating adolescent brain development through sleep heart rate monitoring.

To verify the stability and reliability of the two dimensions, the study also conducted 1000 bootstrap resampling analyses and confirmed that all sleep and brain indicators constituting the two dimensions had consistent significance, ensuring the scientificity and reproducibility of the research results.

Core Result 2: Three sleep-brain biotypes, showing clear gradient differences in sleep, brain and cognition

Based on the above two sleep-brain correlation dimensions, the research team divided adolescents into three significantly different sleep-brain biotypes through hierarchical cluster analysis. Each biotype presents a highly consistent gradient pattern in sleep characteristics, brain structure/function and cognitive performance, and the biotypes have clear quantitative distribution characteristics in the discovery sample: Type 1: 1252 people (38.9%), Type 2: 784 people (24.3%), Type 3: 1186 people (36.8%). The differences among the three biotypes all reached statistically significant levels (FDR-corrected p<0.05).

01、Type 1: Unhealthiest sleep characteristics, with significant disadvantages in brain and cognitive development

This biotype represents a typical poor-sleep group. Its core sleep performance is as follows: the durations of all sleep stages (light sleep, deep sleep, REM sleep) are the shortest among the three groups, bedtime is the latest and wake-up time is the earliest, heart rate at all sleep stages is the highest among the three groups. Meanwhile, the number of nocturnal awakenings is higher and sleep fragmentation is severe, resulting in extremely poor overall sleep quality.

Correspondingly, adolescents of Type 1 have the smallest overall brain volume and the weakest overall cortico-subcortical functional connectivity strength. In terms of cognitive performance, their core cognitive indicators such as crystallized intelligence, picture vocabulary, oral reading recognition and flanker inhibitory control are the lowest among the three groups, with highly significant statistical differences compared with Type 2 and Type 3. They show obvious deficiencies in key cognitive dimensions such as attention, language comprehension and reading ability.

02、Type 2: Moderate sleep characteristics, brain and cognitive development at an intermediate transitional level

This biotype is the group with average sleep characteristics. All sleep indicators (bedtime, sleep duration, sleep heart rate, number of awakenings, etc.) are close to the average level of the study sample without obvious abnormal sleep performance, serving as the "intermediate type" group in sleep characteristics.

Its brain volume and brain functional connectivity indicators are between Type 1 and Type 3, and cognitive performance is also at an intermediate transitional level: it is significantly better than Type 1 only in indicators such as crystallized intelligence, reading recognition and attention, with no statistically significant difference from Type 3. It is the core potential group for sleep intervention and cognitive improvement.

03、Type 3: Optimal sleep characteristics, with sustained and significant advantages in brain and cognitive development

This biotype represents the healthy sleep group. Its core sleep performance is as follows: the durations of all sleep stages are the longest among the three groups, bedtime is the earliest with good sleep continuity, heart rate at all sleep stages is the lowest among the three groups, the number of nocturnal awakenings is small and sleep fragmentation is low, resulting in the best overall sleep quality.

Correspondingly, adolescents of Type 3 have the largest overall brain volume and the strongest overall cortico-subcortical functional connectivity strength, with particularly prominent volume advantages in key brain regions such as the hippocampus and amygdala. In terms of cognitive performance, its core indicators such as crystallized intelligence, vocabulary comprehension, reading ability and attention are significantly better than Type 1, making it the group with optimal cognitive development among the three types, and this cognitive advantage has clear statistical significance.

It is worth noting that the study found no significant differences in academic performance among the three biotypes. The research team speculated that this may be because academic performance is affected by many external factors such as family education environment, learning habits and learning methods. Sleep regulation on the brain acts more directly on basic cognitive ability, serving as an important "underlying support" for academic performance rather than a direct determinant.

Core Result 3: Biotype differences have long-term stability; longitudinal tracking confirms the lasting impact of sleep on brain development

To clarify the stability of the three biotypes and the long-term effect of sleep on brain development, the research team carried out a 4-year longitudinal follow-up analysis on adolescents aged 9-14. From baseline (9-10 years old), 2-year follow-up (11-12 years old) to 4-year follow-up (13-14 years old), the dynamic changes of brain structure, function and cognitive development among the three biotypes were systematically evaluated. Two key conclusions were drawn, further confirming the long-term shaping effect of sleep on adolescent brain development.

Figure 3: Three biotypes with different sleep characteristics and cognitive performance

 (A) Hierarchical clustering based on brain canonical variables reveals three biotypes of adolescents. The heatmap and dendrogram show the hierarchical clustering results of 3,222 individuals (rows) in the discovery sample across two dimensions (columns), using the Euclidean distance between the two dimensions of brain canonical variables (dendrogram). The height of each connecting line in the dendrogram represents the distance between clusters connected by that line.

(B) Sleep patterns of the three biotypes. After adjusting for participants' age and gender, the average z-statistics of the three biotypes across all 18 sleep features were calculated. Blue lines and bar charts represent Biotype 1 (n=1,252), green represents Biotype 2 (n=784), and orange represents Biotype 3 (n=1,186).

(C) Intergroup differences in cognitive performance among the three biotypes. For cognitive domains, only results with statistically significant main effects of biotype in the analysis of covariance (p<0.05, FDR-corrected) are presented. The y-axis denotes unadjusted cognitive performance scores at age 11-12 (consistent with the year when sleep indicators were recorded by devices). The numbers between any two groups represent raw p-values. Paired comparisons were performed using Student's t-test in post-hoc analysis, *p<0.05 after FDR correction. Statistical significance was adjusted for age, gender, body mass index (BMI), pubertal development, family income, parental education level and ethnicity.

1. Gradient differences in cognitive ability remain stable with no time interaction effect

For adolescents of all three biotypes, cognitive abilities including crystallized intelligence, picture vocabulary, reading recognition and attention naturally improve with age, but the cognitive gradient difference of "Type 3＞Type 2＞Type 1" remains stable throughout the 4-year follow-up period.

This means that the cognitive development differences shaped by sleep characteristics are not a short-term phenomenon, but persist in early adolescence and will not disappear naturally with aging. The research team emphasized that this result highlights the importance of early sleep intervention — optimizing sleep habits during the critical period of brain development aged 9-14 can lay a solid foundation for adolescents' cognitive development.

2. Differences in brain structure and function persist, consistent with normal adolescent developmental patterns

At the brain volume level, all three biotypes show the characteristics of slightly decreased cortical volume and increased subcortical volume with age, which is highly consistent with the normal developmental rule of adolescent brain "cortical pruning and subcortical development". However, the brain volume gradient difference of "Type 3＞Type 2＞Type 1" exists significantly at all three time points and does not disappear with age. In particular, volume differences in key brain regions such as the hippocampus and amygdala remain stable.

At the functional connectivity level, the gradient difference in cortico-subcortical functional connectivity remains stable, while biotype differences in intracortical functional connectivity gradually disappear at the age of 13-14. This suggests that sleep exerts dynamic influences on different brain functional networks at different adolescent stages. Cross-regional connections between subcortex and cortex are more sensitive to sleep characteristics and serve as the core target of sleep regulating brain development.

Longitudinal analysis further confirms from the temporal dimension that the shaping effect of sleep characteristics on brain development is long-term and stable. Early sleep differences will translate into sustained differences in brain structure and function, thereby affecting cognitive development trajectories. It provides direct and powerful empirical evidence for "sleep is a key regulatory factor for adolescent brain development".

Figure 4: Correlation analysis of brain volume and functional network connectivity patterns with sleep dimensions.Each sleep dimension is associated with decreased brain volume as well as weakened or enhanced functional network connectivity.

Figure 5: Longitudinal changes in cognitive performance and developmental status of brain volume and functional network connectivity in three adolescent biotypes.

Core Result 4: Dual cross-sample replication verification; research conclusions feature high robustness and universality

To ensure the reliability, generalizability and clinical application value of the research results, the research team completed dual replication verification of core results in two independent follow-up samples, further confirming the robustness of the two sleep-brain correlation dimensions and three biotypes:

Figure 6: Replication verification of sleep-brain dimensions and three sleep patterns.All experimental procedures in (A) and (B) were replicated in the 4-year follow-up sample of 1,190 participants.

(A) The first two brain dimensions in the replication sample are significantly correlated with the first two sleep dimensions respectively. (B) Hierarchical cluster analysis of the first two brain dimensions reveals three adolescent biotypes, each exhibiting unique sleep patterns.

1. In the full 4-year follow-up sample of 1190 adolescents aged 13-14, the two sleep-brain correlation dimensions were successfully replicated. Hierarchical clustering still identified three biotypes whose sleep characteristics were highly similar to those in the discovery sample. Only the sleep heart rate of Type 2 was slightly higher than the average level, and all other core features were completely consistent;

2. In an independent non-overlapping follow-up sample of 596 adolescents, the core "fall asleep time-total duration" dimension still existed significantly. Cluster analysis based on this dimension also identified three sleep biotypes, and their sleep characteristics and cognitive gradients were highly consistent with the discovery sample;

3. The sleep indicator loadings and brain indicator loadings of the discovery sample and validation samples are highly correlated, confirming the universality of the research results among adolescent groups of different age groups.

Clinical Value and Industrial Implications of the Study: Providing Precise Targets for Sleep Intervention and Medical Device R&D

This large-sample, objective-data, longitudinal-tracking study not only reveals the internal mechanism between sleep and adolescent brain development at the basic neuroscience level, but also carries important and far-reaching value at the level of clinical application and medical device industry, providing precise targets for adolescent sleep health management and innovative R&D of sleep-related medical devices.

(1) Providing precise and operable scientific guidance for adolescent sleep intervention

The study clarifies four core sleep factors for promoting adolescent brain development and cognitive improvement: going to bed early, sufficient sleep duration, reducing sleep heart rate and lowering sleep fragmentation. It provides clear scientific basis for families, schools and pediatric clinics to formulate sleep guidance plans for adolescents. Clinicians can develop personalized sleep intervention programs for adolescents based on these four factors, especially for the Type 1 poor-sleep group, to realize early screening, early intervention and early adjustment.

(2) Highlighting the clinical value and application potential of wearable sleep monitoring devices

Objective sleep data were collected by wearable devices throughout this study. Compared with traditional subjective sleep reports, wearable devices can continuously, objectively and accurately capture multi-dimensional sleep indicators such as bedtime, sleep duration, sleep heart rate, sleep stages and awakening times, and can reflect subjects' sleep status more truthfully. They are ideal tools for adolescent sleep biotype screening, long-term sleep monitoring and intervention effect evaluation. This research result further verifies the core value of wearable medical devices in sleep medicine research and clinical application, and provides solid scientific support for the clinical promotion of wearable sleep monitoring devices.

(3) Opening up a new direction for the R&D of sleep-related medical devices

The two sleep-brain correlation dimensions, core affected brain regions and functional networks revealed by the study provide precise scientific targets for the R&D of medical devices for sleep monitoring and auxiliary brain function improvement:

◼ For core indicators such as sleep heart rate and light sleep duration, more accurate wearable monitoring devices can be developed to realize refined assessment of sleep health;

◼ Based on the core target of sensorimotor network-subcortical region connectivity, targeted sleep intervention devices can be developed, such as transcutaneous vagus nerve electrical stimulation, transcranial magnetic stimulation, acoustic and light rhythm intervention and other devices, to improve sleep quality by regulating brain functional connectivity;

◼ Combined with the rule of sleep-brain biotyping, a sleep biotype screening system can be developed to quickly identify adolescent sleep biotypes through wearable device data, providing basis for personalized sleep intervention.

Study Limitations and Future Research Directions

Although this study has achieved breakthrough results, the research team also objectively pointed out certain limitations of the study, which also point out directions for subsequent research and exploration:

1. The study did not incorporate indicators such as sleep regularity, sleep environment and pre-sleep behaviors, failing to fully characterize adolescents' sleep patterns. More sleep-related indicators can be included in the future to achieve more comprehensive analysis of sleep characteristics;

2. The brain network partition adopted the adult-standard Gordon network, lacking a dedicated brain network partition scheme tailored to the characteristics of adolescent brain development. It is necessary to develop brain network partitions adapted to adolescent brains in the future to further improve analysis accuracy;

3. The study did not integrate genomics and epigenetics data, and the genetic mechanism of sleep-brain biotyping has not been explored. In the future, multi-omics data can be combined to deeply analyze the molecular mechanism by which sleep affects brain development, providing more precise molecular targets for sleep intervention;

4. No interventional experiments were carried out in this study. Randomized controlled trials can be conducted in the future to verify the effectiveness of optimizing sleep characteristics in improving adolescent brain development and cognitive ability, providing more direct basis for clinical intervention.