The current research adopted a cross-sectional design, which comprised a sample of adolescents. All validated participants (n=7,246; mean age [standard deviation]=17.98 [1.03]) in Luzhou, Sichuan province, China, from October to December in 2020.

The study was approved by the ethics committee of the affiliated hospital of southwest medical university (SWMUFY-220), and all recruits provided written consent before they answered the questionnaires. Participants were firstly asked to complete a checklist of demographic information (including age, sex, myopia, only-child, twins, left behind, and residence), subsequently applying the IGDS9-SF and BSMAS to identify students whether they addicted and grouped. By using three scales (PHQ-9, GAD-7, and PSQI) to assess the mood and sleep problems. Based on the permission and support of local educational government, the data was collected through online psychological assessments. Invalidated responses were excluded due to providing incomplete data, then responses from a total of 7,246 participants were included in further analysis, and which valid rate of response finally was up to 98.4 %.

This study’s cutoff thresholds of IGDS9-SF and BSMAS refer to previous studies [28,29]. Participants were divided into three groups, IGD group (score of IGDS9-SF ≥21 and BSMAS <9), SMA group (score of BSMAS ≥19 and IGDS9-SF <21), and the control group (score of BSMAS <19 and IGDS9-SF <21). Based on the purpose of this study to compare the psychological characteristics of two different behavioral addictions, 258 participants who showed IGDS9-SF ≥21 and BSMAS ≥19 were excluded in the follow-up analysis, The characteristics of the three groups are described in Table 1. The scores of PHQ-9, GAD-7, and PSQI scales in the three groups were analyzed by t-test.

The network analysis conducted on the three groups involved the utilization of R software (version 4.2.2; https://www.r-project.org/), and the methodology closely followed previously established approaches as outlined in studies [30-33]. Each node within the network represents a specific symptom, and the edges connecting these nodes signify the relationships between the symptoms [33]. To evaluate the associations among IGD, SMA, depression, anxiety, and sleep quality within the network, a sophisticated approach was adopted. Specifically, a sparse Graphical Gaussian Model was employed, which is a statistical model capturing the conditional independence relationships between variables. This model was augmented with the graphical Least Absolute Shrinkage and Selection Operator (LASSO) method, a regularization technique that aids in estimating a sparse network by selecting the most relevant connections while setting others to zero. The detailed implementation of this combined methodology is outlined in the present study [34]. In order to characterize the overall structure of the network, various centrality measures were computed. These included betweenness, indicating the degree of connectivity and the frequency with which a symptom acts as a bridge between other symptoms; closeness, representing the distance centrality and measuring how easily information can flow from one symptom to all others; and strength, reflecting the degree centrality and indicating the number of direct connections a symptom has with others within the network. The computation of these centrality measures provides insights into the importance and influence of specific symptoms within the broader network structure [30,35].

We show the descriptive statistics and groups comparison for the study variables in Table 1. The current study includes 7,246 individuals divided into three groups (IGD, SMA, and control) based on the cutoff thresholds of IGDS9-SF and BSMAS. In the IGD group (n=789), 37.0% are male, with a mean age of 18.0. Meanwhile, non-only child and non-left behind children are 87.5% and 74.1% , respectively. In the SMA group (n=713), the mean age is 16.6, including 82.9% of females. Most individuals in the SMA group are not only-child (84.7%), not left-behind (66.9%). The control group consisted of 5,744 participants, with a mean age of 18.0, including 82.2% females. Most individuals in the control group are not only-child (88.9%), not left-behind (92.6%). The proportion residing in villages is 47.0%, 35.3%, and 48.3% for the IGD, SMA, and control groups, respectively.

As shown in Table 2, compared with the control group, the scores of PHQ-9, GAD-7, and PSQI of adolescents with IGD and SMA were significantly higher (p<0.001) and indicated that adolescents with IGD and SMA have more severe mood and sleep problems.

The total number of edges is 703 in the three networks. The number of non-zero edges was 171, 212, and 349 in the SMA, IGD, and control groups, respectively. As shown in Figure 1, although the overall structure is similar, nodes’ connection mode and the edges’ strength are different among the three networks.

In the SMA group, the strength of PHQ2 “feeling down, depressed, or hopeless” is highest (Figure 2). PHQ3, “trouble falling or staying asleep or sleeping too much,” displays the highest value of closeness and betweenness (Figure 2). In the IGD group, values of strength and closeness of PHQ4 “feeling tired or having little energy” are higher than other symptoms, PSQI7 “daytime dysfunction” shows the highest value of betweenness (Figure 2). In the control group, the values of all three centrality indexes of PSQI7 “daytime dysfunction” are highest (Figure 2), which indicated that “daytime dysfunction” not only had a significant and rapid effect on other symptoms but also was highly correlated with the entire symptom network.

The result of edge weight bootstrapping (Supplementary Figure 1 in the online-only Data Supplement) shows that different groups of networks have certain accuracy. The centrality stability coefficient (Figure 3) shows that the centrality index of the SMA group was 0.75 (strength), 0.75 (closeness), and 0.05 (betweenness). The IGD group was 0.67 (strength), 0.13 (closeness), and 0.05 (betweenness). The control groups were 0.75 (strength), 0.75 (closeness), and 0.36 (betweenness). The results of the poor node centrality test are shown in Supplementary Figure 2 (in the online-only Data Supplement).

The network differences among the three groups were compared. Global differences were found between networks among the three groups (p_control-IGD=0.004, p_control-SMA<0.001, p_IGD-SMA< 0.001), which indicated that the network structure of the three groups is different.

The overall intensity invariance test showed that the intensity of the anxiety-depression-addiction-sleep disorder network was significantly different in three groups (control group=16.06, IGD group=12.76, SMA group=12.03, p_control-IGD<0.01, p_control-SMA<0.01, p_IGD-SMA=0.159).

Compared with the control group, the IGD group showed a statistically significant difference in edge GAD6-IGD2 (p=0.013), PHQ6-IGD4 (p=0.013), PHQ4-IGD5 (p=0.024), and PHQ4-IGD2 (p=0.042) of the anxiety-depression-addiction-sleep disorder symptoms network (Supplementary Table 1 in the online-only Data Supplement). Compared with the control group, the SMA group showed a statistically significant difference in edge SMA2-PSQI2 (p=0.001) and edge SMA5-PSQI5 (p=0.030) of the anxiety-depression-addiction-sleep disorder symptoms network (Supplementary Table 2 in the online-only Data Supplement). The networks of the symptoms in the IGD group and SMA group were compared, and it found the influential edges in IGD were GAD1 (p=0.022), GAD2 (p<0.001), GAD5 (p=0.018), GAD6 (p<0.001), PHQ3 (p=0.006), and PHQ4 (p=0.007), while in SMA were GAD7 (p<0.029) and PHQ7 (p<0.001) (Supplementary Table 3 in the online-only Data Supplement).

Utilizing network analysis, the core symptoms of IA were found to vary between subtypes (SMA and IGD). In the SMA group, “feeling down, depressed, or hopeless” (PHQ2) emerged as a central symptom, consistent with previous findings highlighting depression as a pivotal symptom triggering other problems, including sleep and mental health [41-43]. Additionally, “Trouble falling or staying asleep or sleeping too much” (PHQ3) not only exhibited the strongest transmission between symptoms but also acted as a bridge symptom in the overall network. This finding correlates with reports indicating a strong association between social media use and depression symptoms [44,45], emphasizing the impact of SMA on sleep patterns, daily rhythms 46 and mental well-being (i.e., depression, anxiety, and stress) [47].

In the IGD group, “feeling tired or having little energy” (PHQ4) and “daytime dysfunction” (PSQI7) emerged as central symptoms. The engagement in online gaming, often spanning different time zones, may lead to disrupted sleep-wake patterns and contribute to sleep deprivation [48]. These core symptoms may also influence other psychiatric disorders such as attention-deficit/hyperactivity disorder, obsessive-compulsive disorder, anxiety, and depression [24].

For the control group, “daytime dysfunction” (PSQI7) exhibited the highest centrality indexes, indicating that sleep problems were a core but mild symptom. Despite not meeting the criteria for SMA or IGD, adolescents in the control group experienced sleep quality issues that could impact their daily lives(i.e.,trouble staying awake while driving, eating meals, or engaging in social activity, keep up enough enthusiasm to get things done) [27].

A significant body of evidence links sleep disorders in younger adults to computer and internet use, with increased internet use associated with shorter sleep duration, delayed bedtimes, longer sleep latencies, and increased daytime tiredness in adolescents [48-50]. Therefore, further investigation is crucial to understanding how to mitigate the adverse effects of internet use on adolescents’ mental health and sleep quality.

In sum, expanding the analysis to the item-level network, our findings illuminated distinct core symptoms between SMA and IGD. Depression emerged as the central symptom for SMA, whereas IGD was characterized by sleep-related problems stemming from engagement in internet games, manifesting as feelings of tiredness and low energy leading to daytime dysfunction. A corroborating study in 2020 bolstered our results, emphasizing the strong association between poor sleep and psychological problems in the context of SMA [51]. Plausible mechanisms include the continuous nature of social media use, necessitating preoccupation and waiting for responses without temporal constraints, potentially leading to heightened disturbances in sleep quality [52]. In contrast, online gaming, often involving collaboration with strangers, may impact sleep patterns differently. While expressing inner feelings might be challenging in a gaming platform, social media platforms offer ease in emotional expression [53], possibly contributing to heightened feelings of depression in SMA.

The study highlighted mood and sleep characteristics in adolescents with IGD and SMA, indicating that adolescents with these conditions experience higher levels of depression, anxiety, and sleep problems compared to the control group. Interestingly, the SMA group exhibited higher scores in PHQ-9, GAD-7, and PSQI than the IGD group, suggesting a higher prevalence of depression, anxiety, and sleep disorders in adolescents with social media addiction.

Examining the edge-level network, the SMA group exhibited a significant disparity compared to the control group, particularly in the edge connecting “feeling the urge to use social media more and more” (SMA2) to “sleep latency” (PSQI2), as indicated in Supplementary Table 3 (in the online-only Data Supplement). Sleep latency refers to “how long it usually takes you to fall asleep each night, and the reply cannot get to sleep within 30 minutes” [27] was influenced by an increased desire to use social media, particularly before bedtime, leading to difficulties falling asleep and delayed bedtime. The potential impact of blue light from screens on circadian rhythms was highlighted as a contributing factor [46]. Additionally, the edge connecting “becoming restless or troubled if you are prohibited from using social media” (SMA5) to “sleep disturbances” (PSQI5) underscored the pivotal role of social media restrictions in contributing to subjective physical problems related to sleep. Previous research supported the positive relationship between SMA severity and sleep disturbances, linking it to circadian clock phase delays and subsequent cognitive and emotional consequences [54].

Comparatively, specific edges between IGD and the control group revealed interconnections in Supplementary Table 2 (in the online-only Data Supplement). For instance, the edge linking “feeling more irritability, anxiety, or even sadness when you try to either reduce or stop your gaming activity” (IGD2) to “becoming easily annoyed or irritable” (GAD6), “systematically failing when trying to control or cease your gaming activity” (IGD4) to “feeling bad about yourself or that you are a failure or have let yourself or your family down” (PHQ6), and the edge connecting “have you lost interests in previous hobbies and other entertainment activities as a result of your engagement with the game” (IGD5) to “feeling tired or having little energy” (PHQ4) highlighted the intricate relationship between gaming-related behaviors and mood symptoms. Echoing previous research, the escape from real-world worries and difficulties into online gaming could lead to heightened negative emotions, such as anxiety, anger, and sadness, ultimately culminating in depression [55,56]. The distinct cognitive reappraisal and expressive suppression patterns observed in individuals with IGD may contribute to the potential development or exacerbation of depression [56].

Despite both IGD and SMA being associated with mood and sleep problems, adolescents with IGD exhibited greater interrelatedness to anxiety, nervousness, worrying, and irritability, as reflected in items such as “feeling nervous, anxious, or on edge” (GAD1), “not being able to stop or control worrying” (GAD2), “being so restless that it is hard to sit still” (GAD5), and “becoming easily annoyed or irritable” (GAD6). Concurrently, depression-related symptoms in IGD comprised “trouble falling or staying asleep, or sleeping too much” (PHQ3) and “feeling tired or having little energy” (PHQ4). Combined with the core symptom of IGD, namely “daytime dysfunction,” the mental and sleep mechanisms in adolescents with IGD predominantly revolved around anxiety leading to poor sleep [40]. Conversely, adolescents with SMA were more prone to exhibit emotional states such as “feeling afraid, as if something awful might happen” (GAD7) and “trouble concentrating on things, such as reading the newspaper or watching television” (PHQ7) during social media use. In SMA, the core symptoms involving psychological and sleep aspects were encapsulated in “feeling down, depressed, or hopeless,” which was intricately linked to “sleep latency and sleep disturbances”.

In essence, the associations between IGD/SMA and sleep appear to be driven by different factors. For SMA, the associations primarily revolve around depression and sleep, whereas IGD is characterized by associations with anxiety and sleep. This finding suggests that SMA and IGD may have different psychopathological mechanisms.Furthermore, the distinctive nature of sleep disorders between these two conditions emphasizes the need for tailored interventions targeting specific mood and sleep problems associated with different subtypes of IA in adolescents.

While this study offers valuable insights into the nuanced connections between IA subtypes and mood and sleep symptoms, certain limitations must be acknowledged. The cross-sectional and group-level nature of the data poses challenges in establishing causal relationships among variables, necessitating future longitudinal and intraindividual network analyses. The relatively small sample sizes for IGD and SMA, along with a non-clinical cohort, may limit the generalizability of findings. The reliance on self-report instruments introduces the potential for social desirability and recall biases. Moreover, the study’s cross-sectional design precludes making causal inferences. Therefore, longitudinal intraindividual network analyses, such as dynamic networks [57], should be conducted on the research of core symptoms of IGD and SMA and the differences between them to further explore the interactions of which.

This pioneering study explored mood and sleep symptoms across SMA and IGD in adolescence, offering significant insights into symptomatology changes across these conditions. The core symptoms differ in SMA and IGD, offering critical insights for tailored clinical interventions. Targeting the symptom of depression characterized by “feeling down, depressed, or hopeless” may prove to be the most effective approach in preventing the onset, progression, and persistence of substance misuse and abuse during adolescence. Considering the treatment of IGD, the symptom “feeling tired or having little energy” should be the primary target. Moreover, recognizing the distinct sleep characteristics between IGD and SMA provides a foundation for developing targeted interventions for diverse mood and sleep problems associated with different IA subtypes in adolescents.