Predictors of smartphone addiction in adolescents with depression: combing the machine learning and moderated mediation model approach

doi:10.21203/rs.3.rs-5007740/v1

Download PDF

Research Article

Predictors of smartphone addiction in adolescents with depression: combing the machine learning and moderated mediation model approach

https://doi.org/10.21203/rs.3.rs-5007740/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Smartphone addiction (SA) significantly impacts the physical and mental health of adolescents, and can further exacerbate existing mental health issues in those with depression. However, fewer studies have focused on the predictors of SA in adolescents with depression. This study employs machine learning methods to identify key risk factors for SA and utilizes the Interpretable SHapley Additive exPlanations (SHAP) method to enhance model interpretability and assess the importance of each risk factor. Additionally, by constructing a mediation moderation model, the interactions between significant risk factors is analyzed. The study included 2,203 adolescents with depression. Machine learning results from three models (random forest, logistic regression, and decision tree) consistently identified emotion-focused coping, rumination, and school bullying as the strongest predictors of SA. Further mediation moderation analyses based on the Interaction of Person-Affect-Cognition-Execution (I-PACE) model revealed that rumination significantly mediated the relationship between school bullying and SA, and emotion-focused coping significantly moderated the relationships between school bullying and both rumination and SA. This is the first study to use machine learning to explore the predictors of SA in depressive adolescents and further analyze the interactions among these predictors. Future interventions for SA in adolescents with depression may benefit from psychotherapy that addresses emotion-focused coping and rumination.

Adolescents with depression

smartphone addiction

machine learning

moderated mediation model

Smartphone addiction (SA) is a growing global problem, and it is particularly prominent in China (Olson et al., 2022). SA is defined as the excessive use of smartphones, with the resulting neglect of social activities and adverse effects on an individual's normal life (Otsuka et al., 2023). Adolescents are more susceptible to SA and its negative effects o than other age groups (Park and Lee, 2022). In China, up to 23.8% of adolescents have SA (Zou et al., 2019). Meanwhile, individuals who are depressed tend to alleviate negative emotions and pass the time by using a smartphone, resulting in a higher risk of SA (Lim et al., 2021). Therefore, adolescents with depression may possess a higher risk of SA compared to healthy adolescents. Excessive smartphone use may lead to impaired physical and mental health, and impaired cognitive functioning (Wacks and Weinstein, 2021), and even further exacerbate depressive symptoms. However, current research has mainly focused on healthy adolescents or adults, and no study has yet focused on SA in depressed adolescents. Therefore, exploring the risk factors of SA and how these factors interact in adolescents with depression is warranted.

1.1 Predictors of SA

Currently, there are various theoretical models that can explain the causes of SA (Dong and Potenza, 2014; Billieux et al., 2015; Brand et al., 2016). Among these theoretical models, the Interaction of Person-Affect-Cognition-Execution (I-PACE) model is most widely cited (Ge et al., 2023; Li et al., 2022; Mehmood et al., 2021). It provides persuasive explanation for SA by explaining its causes from different aspects based on past theoretical models (Brand et al., 2016). The I-PACE model suggests that SA is the result of a combination of susceptibility factors, mediators, and moderators, where mediators and moderators play a mediating and moderating role between susceptibility factors and SA, respectively. Based on this theory, this study will recruit and analyze the predictors of SA in adolescents with depressions as many as possible.

1.1.1 Predisposing factors

In the I-PACE model, predisposing factors are used to explain what social and personal characteristics are associated with a higher susceptibility to SA (Brand et al., 2016). Individuals who have experienced stressful life events, such as childhood trauma and school bullying, are more likely to suffer from SA (Geng et al., 2022; Liu et al., 2020). Individuals who lack social support or feel isolated have a tendency toward SA (Hu and Xiang, 2022; Ding et al, 2022). In addition to factors from the external environment, psychopathological characteristics such as depression, anxiety, and sleep problems also contribute to an individual's excessive smartphone use (Zhou et al., 2021; Zhang et al., 2023; Pan et al., 2021). Furthermore, in terms of personality factors, low self-esteem, low psychological resilience, and high levels of alexithymia are also risk factors for SA (Casale et al., 2022; Shi et al., 2023; Lv et al., 2022).

Predisposing factors do not fully explain SA in individuals. Further exploration of the pathways from predisposing factors to SA, i.e., the roles of mediators and moderators in this context, could contribute to comprehensively understanding the mechanism of SA.

1.1.2 Mediating factors

The I-PACE model postulates that individual characteristics can indirectly shape SA behavior through emotional and cognitive features (Brand et al., 2016). For example, depression may indirectly influence SA behavior through emotional and cognitive features like rumination and emotion regulation strategies (Wang et al., 2022; Arrivillaga et al., 2023). Perceived stress is another cognitive factor for addictive behavior, potentially contributing to addiction relapse (Wang et al., 2021; Zhao et al., 2022).

1.1.3 Moderating factors

In the I-PACE model, coping strategies are construed as moderating factors in the context of SA, particularly concerning stressful or emotional problems (Brand et al., 2016). Coping refers to the process individuals manage and restore balance to their mental well-being after cognitive evaluation of urgent environmental or life circumstance. Coping strategies are broadly classified into problem-focused coping, employed to address and resolve difficulties, and emotion-focused coping, implemented to alleviate emotional distress (Folkman et al., 1986). The emotion-focused coping is employed more frequently in people with depression (Kim et al., 2007). It is worth noting that problem-focused coping is negatively associated with SA, while emotion-focused coping is positively associated with SA (Sun et al., 2019).

1.1.4 Demographics

Moreover, we also considered demographic variables that have a significant impact on SA, such as age and gender. Smartphone use is the highest in adolescents and negatively associated with age (Nahas et al., 2018). Additionally, gender differences in SA have been demonstrated (Chan et al., 2021). Women have higher odds and experience more severe cases than men (Davey et al., 2020; Zhao et al., 2023).

1.2 Machine learning

While most previous studies have relied on traditional hypothesis testing methods to identify predictors of SA, the increasing number of variables presents an increasing challenge in comprehensively understand how these variables rank as risk or protective factors, and in discerning their intricate interactions. The extent to which crucial variables contributing to SA remains unclear.

In recent years, machine learning has excelled in processing high-dimensional features and has found widespread application in addiction studies (Mak et al., 2019), among which a few studies investigating SA behavior (Arora et al., 2023; Arpaci, 2023; Beluli, 2023; Elhai et al, 2020; Lee and Kim, 2021; Lee et al., 2021; Giraldo-Jiménez et al., 2022; Mazumdar et al., 2020; Xiao et al., 2023). Unfortunately, the application of machine learning methods to predict SA has several limitations in past research. Firstly, the majority of studies have primarily targeted adults or healthy adolescents, neglecting the group of adolescents with depression, who may have a higher risk of SA. In addition, existing research often adopts a data-driven approach to identify factors, lacking a robust theoretical foundation. Lastly, previous studies have not further explored the interactions between important predictive factors, despite recognizing their interrelatedness, which is essential for developing effective intervention strategies.

1.3 Aim

Taking together, no studies have utilized machine learning methods to explore significant predictors of SA in adolescents with depression. Moreover, the intricate interactions between these critical predictors remains unclear. To address the aforementioned gaps in the literature, the first aim is to utilize a machine learning approach to identify significant predictors of SA among adolescents with depression. To explain the extent to which each factor contributes to the predicted outcome, the Interpretable SHapley Additive exPlanations (SHAP) approach will be incorporated. Building on the theoretical framework provided by the I-PACE model, which has proven instrumental in identifying predictors and their interactions in SA, the second aim of this study is to explore the interactions between the crucial predictors by constructing a moderated mediation model.

2.1 Participants

Participants were adolescents with depression who participated in the Chinese Cohort Study of Adolescent Depression (CSCAD). The project was conducted from March 1, 2021, to January 20, 2022, at 14 study centers (including mental health centers and general hospitals) in 11 provinces in China, to assess the determinants of psychological distress and cognitive function in Chinese adolescents with depression. Inclusion criteria for patients were: (1) unipolar depression or bipolar disorders during depressive episodes, diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-Ⅴ); (2) age 12–18 years; (3) total years of education ≥ 6 years; and (4) right-handedness. Exclusion criteria were: (1) comorbidity of serious physical, infectious, and immune system diseases; (2) presence of traumatic brain injury, epilepsy, or other serious neurological or organic brain disease; and (3) history of severe mental disorders, such as schizophrenia or mental retardation. All participants completed the questionnaire in hospital with the help of a psychiatry or psychology graduate student or psychiatrist.

A total of 2343 adolescents with depression were recruited for this study. After excluding 140 questionnaires with missing data, we obtained 2203 valid questionnaires (effective response rate: 94.15%).

This study was approved by the Ethics Committee of Shenzhen Kangning Hospital (no. : 2020-k021-02), in accordance with the Declaration of Helsinki. All participants and their guardians signed an informed consent form.

2.2 Measures

2.2.1 Smartphone addiction (SA)

SA was assessed with the Mobile Phone Addiction Index (MPAI; Leung, 2008; Huang et al., 2014). The scale consists of 17 items divided into four factors: inability to control carving, feeling anxious and lost, withdrawal or escape, and productivity loss. The items are rated on a five-point scale ranging from 1 (never) to 5 (always). Higher scores indicate higher levels of SA. A total score greater than or equal to 51 was considered a SA. In the present study, Cronbach’s α was 0.89.

2.2.2 Depression

Depression was measured with the Patient Health Questionnaire-9 (PHQ-9, Kroenke et al., 2001). Structured as a nine-item questionnaire, the PHQ-9 incorporates criteria from the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) for major depressive disorder. Respondents rate the frequency of depressive symptoms experienced over the preceding two weeks on a scale ranging from ‘0’ (not at all) to ‘3’ (nearly every day), yielding a total score that corresponds to the severity of depressive symptomatology.

2.2.3 Anxiety

Anxiety was measured by the Generalized Anxiety Disorder (GAD-7, Spitzer et al., 2006). The scale consists of seven items, aligning with diagnostic criteria for generalized anxiety disorder outlined in the DSM-IV. Respondents are asked to rate the frequency of specific symptoms of anxiety over the previous 2 weeks on a scale of ‘0’ (not at all) to ‘3’ (nearly every day). The total score ranges from 0–27, with the higher the total score, the more severe the anxiety.

2.2.4 Sleep problems

The Pittsburgh Sleep Quality Index (PSQI) was used to measure sleep problems (Buysse et al., 1989; Liu et al., 1996). The PSQI consists of 19 items that can be categorized into 7 subscales. The total PSQI score ranges from 0–21, with higher scores indicating more severe sleep problems.

2.2.5 Stress perception

Perceived stress was measured using the Perceived Stress Scale (PSS; Cohen, 1988). The scale consists of 4 items, each rated on a 5-point Likert scale ranging from 1 (never) to 5 (very often). The total score, obtained by summing the scores of all items, reflects the individual's subjective perception of stress. Higher scores indicate higher perceived stress levels.

2.2.6 Loneliness

Loneliness was assessed using the University of California at Los Angeles Loneliness Scale (UCLA-20; Russell et al., 1978; Wang, 1995). The scale consists of 20 entries on a 4-point scale. A higher total score indicates a higher level of loneliness.

2.2.7 Alexithymia

The Toronto Alexithymia Scale (TAS)-20 was used to assess alexithymia (Babgy et al., 1994). The TAS-20 consists of 20 items, each assessing various facets of alexithymia, including difficulty identifying feelings, difficulty describing feelings, and externally-oriented thinking. A 5-point Likert scale was used, with a higher total score indicating a greater degree of alexithymic traits.

2.2.8 Self-esteem

Self-esteem was measured using the Rosenberg Self-Esteem Scale (Rosenberg, 1965). The scale consists of 10 items rated on a 4-point scale, with higher total scores indicating higher self-esteem.

2.2.9 Rumination

The Ruminative Responses Scale (RRS, Nolen-Hoeksema and Morrow, 1991; Han and Yang, 2009) was used to measure how much a person tends to engage in rumination, which involves repeatedly thinking about distressing feelings and their causes. The RRS included three subscales capturing different facets of the ruminative process: reflection, brooding, and depressive rumination. A 4-point scale consisting of 22 items was employed, where a higher total score indicates more severe rumination.

2.2.10 Resilience

Resilience was measured using the 10-item Connor-Davidson resilience Scale (Wang et al., 2010). The scale comprises 10 items rated on a 5-point scale ranging from 0 to 4. Higher scores indicate higher levels of mental toughness.

2.2.11 Coping

Coping was assessed using the Coping Styles Scale for Middle School Students (CSS; Cheng et al., 2000). The scale comprises 36 items and encompasses two distinct subscales: a problem-oriented coping subscale and an emotion-oriented coping subscale. The problem-oriented coping subscale includes three dimensions: problem-solving, seeking social support, and positive rationalization. Meanwhile, the emotion-oriented coping subscale comprises four dimensions: endurance, avoidance, emotional outburst, and fantasy denial. A higher subscale score indicates a greater propensity for the corresponding coping strategy.

2.2.12 Emotion Regulation Strategies

Emotion regulation strategies were measured using the Emotion Regulation Questionnaire (ERQ, Gross and John, 2003; Wang et al., 2007). The questionnaire consists of 10 items, including two emotion regulation strategies: cognitive reappraisal and expressive inhibition. The questionnaire was scored on a five-point Likert scale (1 for "completely disagree"; 5 for "completely agree").

2.2.13 Social support

Social support was measured by the Perceived Social Support Questionnaire (Zimet et al., 1990). The scale consists of 12 items divided into 3 dimensions, including family support, friend support, and support from others. A 7-point scale was employed, where a higher total score reflects a greater perception of social support.

2.2.14 Childhood trauma

Childhood trauma was assessed using the Childhood Trauma Questionnaire (CTQ, Bernstein et al., 2003; Zhao et al., 2005). The questionnaire contains 28 items categorized into 5 dimensions. A 5-point Likert scale was used to score the items, where higher total scores signify more severe childhood trauma.

2.2.15 School bullying

Bullying victimization was measured with the Multidimensional Peer Victimization Scale (MPVS, Mynard and Joseph, 2000). The scale was divided into four dimensions encompassing 16 items: physical victimization, social manipulation, verbal victimization, and attacks on property.

2.3 Statistical and machine learning analysis

Prior to data analysis, in order to rectify a slight sample imbalance in the dataset (labeled 0 : 1503, labeled 1 = 700), we applied the Synthetic Minority Over-sampling Technique (SMOTE) algorithm. In this study, we utilized three classical machine learning models—including Random Forest, Logistic Regression, and Decision Tree—to classify datasets. These three models have demonstrated good classification performance. We maintain a consistent training and prediction process for each model. Initially, we use a grid search parameter optimization method to determine the optimal hyperparameters for each model,. the hyperparameter search range of each model is shown in Table S1. Subsequently, based on the model with the best hyperparameter, we perform a 10-fold cross-validation, using accuracy as the evaluation index to get the optimal prediction score.

While machine learning models frequently excel in classification tasks, they struggle with interpretability(Guidotti et al., 2018). Understanding how each feature affects the model's decision is challenging. Thus, we adopt the SHAP method derived from game theory to interpret machine learning models in a unified way(Lundberg and Lee, 2017). The essence lies in calculating the marginal contribution of features to the model output, and interpreting the "black-box models" from both global and local levels. Our primary findings encompass the top five predictors of SA identified by each machine learning model, focusing on common primary features.

We use the open source scikit_learn library in Python to implement random forest, logistic regression, decision tree model, lattice search algorithm and 10-fold cross-validation. The SMOTE method was implemented using open source imbalanced-learn library, while the SHAP method was implemented using open source shap library in Python.

Subsequently, we explored the interaction between significant predictor variables of SA by constructing a moderated mediation model. The data were standardized, and the SPSS PROCESS was employed to test the moderated mediation model. In PROCESS 3.3, model 4 calculated the mediation model, and model 8 calculated the moderated mediation model. Indirect effects were calculated using a bias-corrected bootstrap procedure. Significance of the mediating effect was determined if the 95% confidence interval (CI) excluded 0.

3.1 Summary Statistics

Table S2 shows the sociodemographic and clinical characteristics of participants. A total of 2203 adolescents with depression were included in this study, with a mean age of 15 years old. Among them, 491 (22.3%) were boys and 1712 (77.7%) were girls. In addition, 758 (34.4%) participants had SA.

3.2 Machine learning

After training, all three models exhibited a prediction accuracy of approximately 70% or higher. Specifically, the average accuracy of 10-fold cross-validation for the Random Forest model reached 77.4%, Logistic Regression achieved a 10-fold cross-validation accuracy of 69.1%, and the Decision Tree model yielded a 10-fold cross-validation accuracy of 71.0%, as shown in Fig. 1

As described above, our aims focused on determining the most consistent and robust predictors of SA across models. The importance of the top five features of each model, as calculated by the SHAP method is shown in Table 1. The results revealed consistent rankings across models, with emotionally oriented coping, rumination, and school bullying consistently occupying the first, second, and third positions, respectively. This consistency underscores the stable significance of these three features in identifying SA across models. Notably, coping with emotions emerged as the foremost feature in all three models. In the Random Forest model, the fourth feature is self-esteem, and the fifth feature is resilience. In the Logistic Regression model, gender and resilience are ranked fourth and fifth, respectively. Lastly, in the Decision Tree model, the fourth characteristic is self-esteem and the fifth is Resilience.

Table 1

Top 5 predictors of smartphone addiction selected by models
	Random forest	Logistic Regression	Decision tree
1	Emotion-focused coping	Emotion-focused coping	Emotion-focused coping
2	Rumination	Rumination	Rumination
3	School bullying	School bullying	School bullying
4	Self-esteem	Gender	Self-esteem
5	Resilience	Resilience	Resilience

3.3 Path analyses

Table 2 illustrates significant correlations between school bullying, rumination, emotion-focused coping, and SA. Specifically, school bullying positively correlated with rumination (r = 0.30, p < 0.001), emotion-focused coping (r = 0.16, p < 0.001), and SA (r = 0.29, p < 0.001). Rumination positively correlated with emotion-focused coping (r = 0.50, p < 0.001) and SA (r = 0.43, p < 0.001). A positive correlation was observed between emotion-focused coping and SA (r = 0.42, p < 0.001).

Table 2

Correlation between variables in path analyses.
	M ± SD	1	2	3	4
1. School bullying	34.46 ± 16.78	1
2. Rumination	62.11 ± 14.41	0.30***	1
3. Emotion-focused coping	45.16 ± 8.50	0.16***	0.50***	1
4. Smartphone addiction	45.31 ± 13.48	0.29***	0.43***	0.42***	1
Notes: p<0.05, p<0.01, **p<0.001.

Next, we examined the mediating role of rumination between school bullying and SA. Controlling for gender and age, school bullying significantly positively predicted both SA (β = .13, t = 7.94, p < .001) and rumination (β = .27, t = 14.92, p < .001); and rumination significantly positively predicted SA (β = .35, t = 18.91, p < .001). A bias-corrected Bootstrap test indicated a significant mediation effect of rumination, with an indirect effect value of .09 and a 95% confidence interval of (.08, .11). This suggests that school bullying have both a direct effect on SA in adolescents with depression and an indirect effect on SA through the partially mediated effect of rumination.

Finally, we explored the moderating role of emotion-focused coping between school bullying and both SA and rumination. As shown in Table 3, after motion-focused coping entered into the model, the product term of school bullying and motion-focused coping emerged as a significant predictor of rumination(β = − .006, t = -3.468, p <. 01) and SA (β = − .004, t = -2.137, p <. 05). This implies that motion-focused coping moderated significantly on the pathway from school bullying to rumination and on the direct pathway from school bullying to SA. The detailed model is shown in Fig. 3.

Table 3

Moderated mediation effect test.
Outcome variable	Predictive variable	R	R²	F	β	t
Rumination	Age	0.548	0.300	189.082***	0.109	0.668
	Sex				0.808	1.291
	School bullying Emotion-focused coping School bullying×Smartphone addiction				0.469 0.978 -0.006	5.882* 14.667* -3.468**
SA	Age	0.521	0.271	136.992***	-0.440	3.718**
	Sex				-1.060	-1.775
	School bullying				0.285	3.718**
	Rumination Emotion-focused coping School bullying×Smartphone addiction				0.229 0.573 -0.004	10.814* 8.583* -2.137*
Notes: All variables in the model have been standardized. SSS refers to subjective socioeconomic status

*p<0.05, **p<0.01, ***p<0.001.

To elucidate the nature of the interaction effect, simple slope analyses were conducted at high and low levels of emotion-focused coping (one standard deviation above versus one standard deviation below the sample mean). As depicted in Fig. 2(a), the positive correlation between school bullying and rumination is stronger when at lower levels of emotion-focused coping, while it attenuates at higher levels. Similarly, in Fig. 2(b), the positive correlation between school bullying and SA is stronger at lower levels of emotion-focused coping and weaker at higher levels.

As far as we know, this study breaks new ground by using machine learning and SHAP explainability methods to analyze the risk factors of SA in adolescents with depression, using a large, nationally representative sample. Furthermore, it explores how these risk factors interact through a moderated mediation model. The findings undescore the prominence of emotion-focused coping, rumination, and school bullying as robust and top determinants of SA, consistently observed among 35 risk factors across diverse models. Notably, through subsequent mediation moderation analysis grounded in the I-PACE model, it is revealed that rumination serves as a significant mediator in the relationship between school bullying and SA, while emotion-focused coping significantly moderates the relationships between school bullying and both rumination and SA. This amalgamation of data-driven approaches and theoretical models not only advances our understanding of SA predictors but also provides a framework for targeted intervention strategies for depressed adolescents.

In recent years, machine learning has garnered increasing attention from researchers due to its ability to identify complex nonlinear relationships between variables and its powerful high-dimensional data processing capabilities. However, traditional machine learning methods cannot provide the contribution of each factor to the prediction outcome. The advent of the SHAP method has addressed this shortcoming. This is the first study to combine machine learning with the SHAP method to investigate the predictive factors of SA in adolescents with depression. SA is influenced by numerous factors, and previous studies have struggled to comprehensively include relevant variables. Guided by the I-PACE model, this study has endeavored to collect as many predisposing factors, mediating factors, moderating factors, and relevant demographic variables as possible. Ultimately, the results from three machine learning models—random forest, logistic regression, and decision tree—consistently indicate that emotion-focused coping, rumination, and school bullying are significant predictive factors for SA in adolescents with depression.

Consistent with previous studies (Vessey et al., 2022; Chen et al., 2021; Kim, 2021), this study found that school bullying significantly predicted SA. Among Chinese middle school students, the prevalence of bullying is between 1.68% and 10.60%, and the victimization rate ranges from 5.91–25.70% (Peng et al., 2019). Individuals who suffer from school bullying are often ostracized by their peers as well, prompting them to hate group life even more (Copeland et al., 2013), which makes them withdrawn and prone to SA. According to the Internet compensation theory, when experiencing psychosocial problems, individuals may use the Internet or smartphones to escape real life issues and relieve negative emotions (Kardefelt-Winther, 2014). Apparently, such a strategy led to poorer well-being and behavioral problems, like SA (Weidman et al., 2012). Previous studies have only explored the direct impact of bullying in schools on SA, while this study further focuses on other factors that may influence this pathway.

For the first time, we found the mediating role of rumination between school bullying and SA in adolescent depression. This means that the effect of school bullying on SA was predicted on one hand through the direct effect of school bullying and on the other hand through the mediating role of rumination. Consistent with the results of previous research, school bullying triggers individuals to rumination. Then to cope with the negative emotions associated with rumination, they overuse smartphones (Zou et al., 2023; Peng et al., 2022). According to the cognitive-behavioral model of pathological Internet use (Davis, 2001), although SA is a result of negative life events such as school bullying, maladaptive cognitions about the self, namely rumination, is the most central factor causing SA directly.

Emotion-focused coping is the most important predictor of SA in adolescents with depression found in our study, across all machine learning models, while problem -focused coping does not predict SA. In depressive patients, emotion-focused coping can lead to increased SA due to its nature of providing immediate emotional relief but failing to address underlying issues. Emotion-focused coping strategies, particularly those involving avoidance, may reinforce negative behavior patterns and worsen depressive symptoms, creating a feedback loop that increases SA. Problem-focused coping, while potentially less immediately gratifying, aims to tackle the root causes of stress and depression, offering long-term benefits and reducing the need for smartphones as a coping mechanism.

The present study found that both the effects of school bullying on SA and school bullying on rumination were moderated by emotion-focused coping in adolescents with depression. The effects of school bullying on both SA and rumination were stronger in depressed adolescents with lower levels of emotion-focused coping compared to those with higher levels of emotion-focused coping (Stanton et al., 1994; Stanton and Low, 2012). Emotion-focused coping may provide immediate, albeit temporary, relief from depressive symptoms through smartphone use. Therefore, for depressed adolescents with lower levels of emotion-focused coping, due to the lack of effective coping methods, the negative cognition brought about by school bullying, namely rumination, and behavioral problems, namely SA, may be more intense. However, in the long run, individuals with higher levels of emotion-focused coping have higher levels of rumination and SA brought about by school bullying than those with lower levels of emotion-focused coping. The method of emotion-focused coping is still a risk factor for SA in adolescent patients with depression.

In conclusion, we use machine learning to explore the predictors of SA in depressed adolescents and to analyze the interactions between important predictors. These findings have important implications for reducing SA in adolescents with depression, especially those school bullying victims. In future clinical work, we can intervene in SA from two perspectives: emotion-directed coping and rumination. First, we guide patients to reduce the use of emotion-directed coping in favor of problem-oriented coping. Second, rumination-focused cognitive behavior therapy, cognitive bias modification, and cognitive control training can be used to reduce rumination (Watkins and Roberts, 2020).

There are some limitations of this study. First, the cross-sectional design limits the ability to establish causal relationship between risk factors and SA in adolescents with depression or to assess changes in variables over time. Further analyses using longitudinal data could be conducted in the future to determine causality. Second, the data were based on subjective self-reports, which may have recall bias or social desirability bias. Future studies could benefit from using more objective measures. Finally, the study sample consisted exclusively of adolescents with depression, resulting in the limited generalizability of the findings. Future research therefore should explore whether these results apply to adults with depression or to populations from diverse cultural backgrounds.

The present study, despite certain limitations, is, to our knowledge, the first study to use a machine learning approach to explore the predictors of SA in adolescents with depression and to explore how important predictors interact with each other to lead to SA, according to the I-PACE model. This study concluded that emotion-focused coping was the strongest predictor of SA in adolescent depression, and rumination thinking and school bullying were significant predictors. It also reveals the potential pathways through which school bullying influences SA in depressed adolescents, providing a framework for the prevention and intervention of SA in depressed adolescents.

Author Contribution

Yongjie Zhou, Chenran Pei: Conception of study design, methodology, data analysis, visualization, writing—original draft preparation; Hailong Yin, Rongting Zhu, Xuankun Zhang, Tian Lan, Junchang Li: Methodology, visualization, data analysis, and data curation; Nan Yan, Lan Wang: resources, writing—review and editing, supervision, Lijuan Huo: Conception of study design, resources, funding acquisition, writing—review and editing, supervision.

Acknowledgement

We are grateful to the adolescents and parents who participated in our study.

Aldwin, C. M., & Revenson, T. A. (1987). Does coping help? A reexamination of the relation between coping and mental health. Journal of Personality and Social Psychology, 53(2), 337.
Arrivillaga, C., Elhai, J. D., Rey, L., & Extremera, N. (2023). Depressive Symptomatology Is Associated With Problematic Smartphone Use Severity in Adolescents: The Mediating Role of Cognitive Emotion Regulation Strategies. Cyberpsychology-Journal of Psychosocial Research on Cyberspace, 17(3), Article 2. https://doi.org/10.5817/cp2023-3-2
Arora, A., Chakraborty, P., Bhatia, M., & Puri, A. (2023). Intelligent Model for Smartphone Addiction Assessment in University Students using Android Application and Smartphone Addiction Scale. International Journal of Education and Management Engineering, 13(1), 29.
Arpaci, I. (2023). Predicting problematic smartphone use based on early maladaptive schemas by using machine learning classification algorithms. Journal of Rational-Emotive & Cognitive-Behavior Therapy, 41(3), 634-643.
Bagby, R. M., Parker, J. D. A., & Taylor, G. J. (1994). THE 20-ITEM TORONTO-ALEXITHYMIA-SCALE .1. ITEM SELECTION AND CROSS-VALIDATION OF THE FACTOR STRUCTURE. Journal of Psychosomatic Research, 38(1), 23-32. https://doi.org/10.1016/0022-3999(94)90005-1
Beluli, A. (2023). Machine Learning-Based Prediction Model for the Measurement of Mobile Addiction. In Designing and Developing Innovative Mobile Applications (pp. 56-66). IGI Global.
Bernstein, D. P., Stein, J. A., Newcomb, M. D., Walker, E., Pogge, D., Ahluvalia, T., Stokes, J., Handelsman, L., Medrano, M., Desmond, D., & Zule, W. (2003). Development and validation of a brief screening version of the Childhood Trauma Questionnaire. Child Abuse & Neglect, 27(2), 169-190. https://doi.org/10.1016/s0145-2134(02)00541-0
Billieux, J., Maurage, P., Lopez-Fernandez, O., Kuss, D. J., & Griffiths, M. D. (2015). Can disordered mobile phone use be considered a behavioral addiction? An update on current evidence and a comprehensive model for future research. Current Addiction Reports, 2(2), 156-162.
Brand, M., Young, K. S., Laier, C., Wolfling, K., & Potenza, M. N. (2016). Integrating psychological and neurobiological considerations regarding the development and maintenance of specific Internet-use disorders: An Interaction of Person-Affect-Cognition-Execution (I-PACE) model. Neuroscience and Biobehavioral Reviews, 71, 252-266. https://doi.org/10.1016/j.neubiorev.2016.08.033
Buysse, D. J., Reynolds, C. F., 3rd, Monk, T. H., Berman, S. R., & Kupfer, D. J. (1989). The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry research, 28(2), 193-213. <Go to ISI>://MEDLINE:2748771
Casale, S., Fioravanti, G., Benucci, S. B., Falone, A., Ricca, V., & Rotella, F. (2022). A meta-analysis on the association between self-esteem and problematic smartphone use. Computers in Human Behavior, 134, Article 107302. https://doi.org/10.1016/j.chb.2022.107302
Chan, D. L., Wong, C. S., Hui, C. L., Chan, S. K., & Lee, E. H. (2021). Smartphone internet addiction among hong kong young adults: The role of gender and depression. European Psychiatry, 64, S211-S211. https://doi.org/10.1192/j.eurpsy.2021.562
Chen, Y. Y., Zhu, J. J., & Zhang, W. (2021). Reciprocal longitudinal relations between peer victimization and mobile phone addiction: The explanatory mechanism of adolescent depression. Journal of Adolescence, 89, 1-9. https://doi.org/10.1016/j.adolescence.2021.03.003
Cheng, S. L., Zheng, Q. Q., Pan, J. N., & Zheng, S. S. (2000). Preliminary Development of Coping Style Scale for Middle School Students. Chinese Journal of Clinical Psychology, (04), 211-214+237
Cohen, S. (1988). Perceived stress in a probability sample of the United States.
Copeland, W. E., Wolke, D., Angold, A., & Costello, E. J. (2013). Adult Psychiatric Outcomes of Bullying and Being Bullied by Peers in Childhood and Adolescence. Jama Psychiatry, 70(4), 419-426. https://doi.org/10.1001/jamapsychiatry.2013.504
Davey, A., Nasser, K., & Davey, S. (2020). Gender differential for smart phone addiction and its predictors among adolescents: Assessing relationship with self control via sem approach. Journal of Indian Association for Child and Adolescent Mental Health, 16(3), 80-101.
Davis, R. A. (2001). A cognitive-behavioral model of pathological Internet use. Computers in Human Behavior, 17(2), 187-195. https://doi.org/10.1016/s0747-5632(00)00041-8
Ding, Y. M., Wan, X., Lu, G. L., Huang, H. T., Liang, Y. P., Yu, J. F., & Chen, C. R. (2022). The associations between smartphone addiction and self-esteem, self-control, and social support among Chinese adolescents: A meta-analysis. Frontiers in Psychology, 13, Article 1029323. https://doi.org/10.3389/fpsyg.2022.1029323
Dong, G. H., & Potenza, M. N. (2014). A cognitive-behavioral model of Internet gaming disorder: Theoretical underpinnings and clinical implications. Journal of Psychiatric Research, 58, 7-11. https://doi.org/10.1016/j.jpsychires.2014.07.005
Elhai, J. D., & Contractor, A. A. (2018). Examining latent classes of smartphone users: Relations with psychopathology and problematic smartphone use. Computers in Human Behavior, 82, 159-166.
Folkman, S., Lazarus, R. S., Gruen, R. J., & DeLongis, A. (1986). Appraisal, coping, health status, and psychological symptoms. Journal of Personality and Social Psychology, 50(3), 571-579.
Ge, J. M., Liu, Y., Cao, W. J., & Zhou, S. Y. (2023). The relationship between anxiety and depression with smartphone addiction among college students: The mediating effect of executive dysfunction. Frontiers in Psychology, 13, Article 1033304.
Geng, J. Y., Bao, L., Wang, H. X., Wang, J., Gao, T., & Lei, L. (2022). Does childhood maltreatment increase the subsequent risk of problematic smartphone use among adolescents? A two-wave longitudinal study. Addictive Behaviors, 129, Article 107250. https://doi.org/10.1016/j.addbeh.2022.107250
Giraldo-Jiménez, C. F., Gaviria-Chavarro, J., Sarria-Paja, M., Bermeo Varón, L. A., Villarejo-Mayor, J. J., & Rodacki, A. L. F. (2022). Smartphones dependency risk analysis using machine-learning predictive models. Scientific Reports, 12(1), 22649.
Gross, J. J., & John, O. P. (2003). Individual differences in two emotion regulation processes: Implications for affect, relationships, and well-being. Journal of Personality and Social Psychology, 85(2), 348-362. https://doi.org/10.1037/0022-3514.85.2.348
Guidotti, R., Monreale, A., Ruggieri, S., Turini, F., Giannotti, F., & Pedreschi, D. (2018). A survey of methods for explaining black box models. ACM computing surveys (CSUR), 51(5), 1-42.
Han, X., Yang, H. F.. (2009). Chinese Version of Nolen-Hoeksema Ruminative Responses Scale(RRS) Used in 912 College Students: Reliability and Validity. Chinese Journal of Clinical Psychology, 17(5), 550-551.
Handing, E. P., Strobl, C., Jiao, Y., Feliciano, L., & Aichele, S. (2022). Predictors of depression among middle-aged and older men and women in Europe: A machine learning approach. Lancet Regional Health-Europe, 18, 100391.
Hu, Z. F., & Xiang, Y. H. (2022). Who Is the Chief Culprit, Loneliness, or Smartphone Addiction? Evidence from Longitudinal Study and Weekly Diary Method. International Journal of Mental Health and Addiction. https://doi.org/10.1007/s11469-022-00892-0
Huang, H., Niu, L. Y., Zhou, C. Y., Wu, H. M. (2014). Realiability and validity of Mobile Phone Addiction Index for Chinese College Students. Chinese Journal of Clinical Psychology, 22(05), 835-838.
Kelly, O., Matheson, K., Ravindran, A., Merali, Z., & Anisman, H. (2007). Rumfnative-coping among patients with dysthymia before and after pharmacotheraty. Depression and Anxiety, 24(4), 233-243. https://doi.org/10.1002/da.20236
Kim, J. H. (2021). Factors Associated with Smartphone Addiction Tendency in Korean Adolescents. International Journal of Environmental Research and Public Health, 18(21), Article 11668. https://doi.org/10.3390/ijerph182111668
Kroenke, K., Spitzer, R. L., & Williams, J. B. W. (2001). The PHQ-9 - Validity of a brief depression severity measure. Journal of General Internal Medicine, 16(9), 606-613. https://doi.org/10.1046/j.1525-1497.2001.016009606.x
Lee, C. P., Beckert, T., & Marsee, I. (2018). Well-being and Substance Use in Emerging Adulthood: The Role of Individual and Family Factors in Childhood and Adolescence. Journal of Child and Family Studies, 27(12), 3853-3865.
Lee, H., Ahn, H., Choi, S., & Choi, W. (2014). The SAMS: Smartphone Addiction Management System and Verification. Journal of Medical Systems, 38(1), Article 1.
Lee, J., & Kim, W. (2021). Prediction of Problematic Smartphone Use: A Machine Learning Approach. International Journal of Environmental Research and Public Health, 18(12), Article 6458. https://doi.org/10.3390/ijerph18126458
Lee, J., & Kim, W. (2021). Prediction of problematic smartphone use: A machine learning approach. International Journal of Environmental Research and Public Health, 18(12), 6458.
Leung, L. (2008). LINKING PSYCHOLOGICAL ATTRIBUTES TO ADDICTION AND IMPROPER USE OF THE MOBILE PHONE AMONG ADOLESCENTS IN HONG KONG. Journal of Children and Media, 2(2), 93-113.
Li, L., Niu, Z. M., Mei, S. L., & Griffiths, M. D. (2022). A network analysis approach to the relationship between fear of missing out (FoMO), smartphone addiction, and social networking site use among a sample of Chinese university students. Computers in Human Behavior, 128, Article 107086.
Lim, P. K., Nordin, A. S. A., Yee, A., & Tan, S. B. (2021). Prevalence of Smartphone Addiction in Patients with Depression and Its Association with Depression Severity: a Cross-sectional Study. International Journal of Mental Health and Addiction, 19(4), 919-933.
Lin, Y. H., Chang, L. R., Lee, Y. H., Tseng, H. W., Kuo, T. B. J., & Chen, S. H. (2014). Development and Validation of the Smartphone Addiction Inventory (SPAI). Plos One, 9(6), Article e98312.
Liu, Q. Q., Yang, X. J., Hu, Y. T., & Zhang, C. Y. (2020). Peer victimization, self-compassion, gender and adolescent mobile phone addiction: Unique and interactive effects. Children and Youth Services Review, 118, Article 105397. https://doi.org/10.1016/j.childyouth.2020.105397
Liu, X. C., Tang, M. Q., Hu L., Wng, A. Z., Wu , H. Z., Zhao, G. F., Gao, C. N., & Li, W. S. (1996). Reliability and validity of the Pittsburgh sleep quality index. Chinese Journal of Psychiatry, 29(2). 103-107.
Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in neural information processing systems, 30.
Lv, J. P., Ren, H., Qin, Z. Y., Hu, Y. Y., Cao, R. L., Liang, L. L., Li, C. N., Meng, C. C., Guo, X. M., Fei, J. S., Mei, H. C., & Mei, S. L. (2022). Alexithymia and Mobile Phone Addiction Among College Students With and Without Siblings: a Moderated Mediation of Depression and Gender. International Journal of Mental Health and Addiction. https://doi.org/10.1007/s11469-022-00761-w
Mak, K. K., Lee, K., & Park, C. (2019). Applications of machine learning in addiction studies: A systematic review. Psychiatry research, 275, 53-60. https://doi.org/10.1016/j.psychres.2019.03.001
Mazumdar, A., Karak, G., & Sharma, S. (2020). MACHINE LEARNING MODEL FOR PREDICTION OF SMARTPHONE ADDICTION. Turkish Journal of Computer and Mathematics Education (TURCOMAT), 11(1), 545-550.
Mehmood, A., Bu, T. Y., Zhao, E. R. Y., Zelenina, V., Alexander, N., Wang, W. T., Siddiqi, S. M., Qiu, X. H., Yang, X. X., Qiao, Z. X., Zhou, J. W., & Yang, Y. J. (2021). Exploration of Psychological Mechanism of Smartphone Addiction Among International Students of China by Selecting the Framework of the I-PACE Model. Frontiers in Psychology, 12, Article 758610.
Mynard, H., & Joseph, S. (2000). Development of the multidimensional peer-victimization scale. Aggressive Behavior, 26(2), 169-178. https://doi.org/10.1002/(sici)1098-2337(2000)26:2<169::Aid-ab3>3.0.Co;2-a
Nahas, M., Hlais, S., Saberian, C., & Antoun, J. (2018). Problematic smartphone use among Lebanese adults aged 18-65 years using MPPUS-10. Computers in Human Behavior, 87, 348-353.
Nolenhoeksema, S., & Morrow, J. (1991). A PROSPECTIVE-STUDY OF DEPRESSION AND POSTTRAUMATIC STRESS SYMPTOMS AFTER A NATURAL DISASTER - THE 1989 LOMA-PRIETA EARTHQUAKE. Journal of Personality and Social Psychology, 61(1), 115-121. https://doi.org/10.1037/0022-3514.61.1.115
Olson, J. A., Sandra, D. A., Colucci, E. S., Al Bikaii, A., Chmoulevitch, D., Nahas, J., Raz, A., & Veissiere, S. P. L. (2022). Smartphone addiction is increasing across the world: A meta-analysis of 24 countries. Computers in Human Behavior, 129, Article 107138.
Otsuka, Y., Kaneita, Y., Itani, O., & Matsumoto, Y. (2023). The Japanese Youth Version of the Smartphone Addiction Scale Among the Youth in Japan: Reliability and Validity Assessment. International Journal of Mental Health and Addiction, 21(1), 292-307.
Pan, W., Chen, X. Y., Ji, F. F., Chen, M., & Lu, J. (2021). Correlation between mobile phone addiction tendency and its related risk factor among Chinese college students: A cross-sectional study. Tropical Journal of Pharmaceutical Research, 20(8), 1729-1737. https://doi.org/10.4314/tjpr.v20i8.26
Papalia, D. E., Olds, S. W., and Feldman, R. D. (2007). Human Development. New York, NY: McGraw-Hill.
Park, Y., & Lee, S. (2022). Gender differences in smartphone addiction and depression among Korean adolescents: Focusing on the internal mechanisms of attention deficit and self-control. Computers in Human Behavior, 136, Article 107400.
Pearson, R., Pisner, D., Meyer, B., Shumake, J., & Beevers, C. G. (2019). A machine learning ensemble to predict treatment outcomes following an Internet intervention for depression. Psychological Medicine, 49(14), 2330-2341.
Peng, Y., Zhou, H. L., Zhang, B., Mao, H. L., Hu, R. T., & Jiang, H. B. (2022). Perceived stress and mobile phone addiction among college students during the 2019 coronavirus disease: The mediating roles of rumination and the moderating role of self-control. Personality and Individual Differences, 185, Article 111222. https://doi.org/10.1016/j.paid.2021.111222
Peng ZW, Ding CX, Chen XJ. Research progress on epidemic characteristics and prevention of campus bullying among middle school students in China. Inj Med (electronic version). 2019;8(03):41–7.
Rosenberg, M. (1965). Society and the adolescent self-image. Social Forces, 3(2).
Russell, D., Peplau, L. A., & Ferguson, M. L. (1978). Developing a measure of loneliness. Journal of personality assessment, 42(3), 290-294. https://doi.org/10.1207/s15327752jpa4203_11
Shi, Z. F., Guan, J. L., Chen, H. H., Liu, C. Z., Ma, J., & Zhou, Z. H. (2023). Teacher-student relationships and smartphone addiction: the roles of achievement goal orientation and psychological resilience. Current Psychology, 42(20), 17074-17086. https://doi.org/10.1007/s12144-022-02902-9
Spitzer, R. L., Kroenke, K., Williams, J. B. W., & Lowe, B. (2006). A brief measure for assessing generalized anxiety disorder - The GAD-7. Archives of Internal Medicine, 166(10), 1092-1097. https://doi.org/10.1001/archinte.166.10.1092
Stanton, A. L., Danoffburg, S., Cameron, C. L., & Ellis, A. P. (1994). Coping through emotional approach problems of conceptualization and confounding. Journal of Personality and Social Psychology, 66(2), 350-362. https://doi.org/10.1037/0022-3514.66.2.350
Stanton, A. L., & Low, C. A. (2012). Expressing Emotions in Stressful Contexts: Benefits, Moderators, and Mechanisms. Current Directions in Psychological Science, 21(2), 124-128. https://doi.org/10.1177/0963721411434978
Sun, J. N., Liu, Q. X., & Yu, S. (2019). Child neglect, psychological abuse and smartphone addiction among Chinese adolescents: The roles of emotional intelligence and coping style. Computers in Human Behavior, 90, 74-83.
Vessey, J. A., Difazio, R. L., Neil, L. K., & Dorste, A. (2022). Is there a relationship between youth bullying and internet addiction? An integrative review. International Journal of Mental Health and Addiction, 1-25.
Wacks, Y., & Weinstein, A. M. (2021). Excessive Smartphone Use Is Associated With Health Problems in Adolescents and Young Adults. Frontiers in Psychiatry, 12, Article 669042.
Wang, D, F. (1995). A study on the reliability and validity of the Russel Loneliness Scale. Chinese Journal of Clinical Psychology, 3(1), 23-25.
Wang, L., Liu, H. C., Li, Z. Q., & Du, W. (2007). Reliability and Validity of Emotion Regulation Questionnaire Chinese Revised Version. China Journal of Health Psychology. 15(06).
Wang, L., Shi, Z. B., Zhang, Y. Q., & Zhang, Z. (2010). Psychometric properties of the 10-item Connor-Davidson Resilience Scale in Chinese earthquake victims. Psychiatry and Clinical Neurosciences, 64(5), 499-504. https://doi.org/10.1111/j.1440-1819.2010.02130.x
Wang, W. B., Mehmood, A., Li, P., Yang, Z. N., Niu, J. B., Chu, H. Y., Qiao, Z. X., Qiu, X. H., Zhou, J. W., Yang, Y. J., & Yang, X. X. (2021). Perceived Stress and Smartphone Addiction in Medical College Students: The Mediating Role of Negative Emotions and the Moderating Role of Psychological Capital. Frontiers in Psychology, 12, Article 660234. https://doi.org/10.3389/fpsyg.2021.660234
Wang, Y., Yang, H. B., Montag, C., & Elhai, J. D. (2022). Boredom proneness and rumination mediate relationships between depression and anxiety with problematic smartphone use severity. Current Psychology, 41(8), 5287-5297. https://doi.org/10.1007/s12144-020-01052-0
Watkins, E., & Brown, R. G. (2002). Rumination and executive function in depression: an experimental study. Journal of Neurology Neurosurgery and Psychiatry, 72(3), 400-402. https://doi.org/10.1136/jnnp.72.3.400
Watkins, E. R., & Roberts, H. (2020). Reflecting on rumination: Consequences, causes, mechanisms and treatment of rumination. Behaviour Research and Therapy, 127, Article 103573. https://doi.org/10.1016/j.brat.2020.103573
Weidman, A. C., Fernandez, K. C., Levinson, C. A., Augustine, A. A., Larsen, R. J., & Rodebaugh, T. L. (2012). Compensatory internet use among individuals higher in social anxiety and its implications for well-being. Personality and Individual Differences, 53(3), 191-195.
Wills, T. A., Sandy, J. M., & Yaeger, A. M. (2001). Time perspective and early-onset substance use: A model based on stress-coping theory. Psychology of Addictive Behaviors, 15(2), 118-125. https://doi.org/10.1037/0893-164x.15.2.118
Xiao, B. W., Parent, N., Rahal, L., & Shapka, J. (2023). Using Machine Learning to Explore the Risk Factors of Problematic Smartphone Use among Canadian Adolescents during COVID-19: The Important Role of Fear of Missing Out (FoMO). Applied Sciences-Basel, 13(8), Article 4970. https://doi.org/10.3390/app13084970
Zanini, D. S., & Forns, M. (2014). Does coping mediate personality and behavioural problems relationship? Psychology, 2014.
Zhang, L., Wang, B. K., Xu, Q., & Fu, C. (2023). The role of boredom proneness and self-control in the association between anxiety and smartphone addiction among college students: a multiple mediation model. Frontiers in Public Health, 11, Article 1201079. https://doi.org/10.3389/fpubh.2023.1201079
Zhao, H., Song, T. J., Rafik-Galea, S., Dong, J. H., Fitriana, M., Ji, Y. H., & Zhang, J. L. (2023). The reciprocal relationships between meaning in life and smartphone addiction among Chinese college students: evidence from a three-wave cross-lagged panel model. Frontiers in Public Health, 11, Article 1202741. https://doi.org/10.3389/fpubh.2023.1202741
Zhao, X. F., Zhang, Y. L., Li, L. F., & Zhou, Y. F. (2005). Reliability and validity of the Chinese version of childhood trauma questionnaire. Chinese Journal of Clinical Rehabilitation, 9(20), 105-107.
Zhao, Z. T., Zhao, S., Wang, Q., Zhang, Y. R., & Chen, C. C. (2022). Effects of Physical Exercise on Mobile Phone Addiction in College Students: The Chain Mediation Effect of Psychological Resilience and Perceived Stress. International Journal of Environmental Research and Public Health, 19(23), Article 15679. https://doi.org/10.3390/ijerph192315679
Zhou, H., Dang, L., Lam, L. W., Zhang, M. X., & Wu, A. M. S. (2021). A cross-lagged panel model for testing the bidirectional relationship between depression and smartphone addiction and the influences of maladaptive metacognition on them in Chinese adolescents. Addictive Behaviors, 120, Article 106978. https://doi.org/10.1016/j.addbeh.2021.106978
Zimet, G. D., Powell, S. S., Farley, G. K., Werkman, S., & Berkoff, K. A. (1990). PSYCHOMETRIC CHARACTERISTICS OF THE MULTIDIMENSIONAL SCALE OF PERCEIVED SOCIAL SUPPORT. Journal of personality assessment, 55(3-4), 610-617. https://doi.org/10.1207/s15327752jpa5503&4_17
Zou, H., Huang, J., Zhang, W., Wu, J., Wu, W., & Huo, L. (2023). The effect of cyberbullying victimization and traditional bullying victimization on suicidal ideation among Chinese female college students: The role of rumination and insomnia. Journal of Affective Disorders, 340, 862-870. https://doi.org/10.1016/j.jad.2023.08.099
Zou, Y. F., Xia, N., Zou, Y. Q., Chen, Z., & Wen, Y. F. (2019). Smartphone addiction may be associated with adolescent hypertension: a cross-sectional study among junior school students in China. Bmc Pediatrics, 19(1), Article 310. https://doi.org/10.1186/s12887-019-1699-9

No competing interests reported.

Supplematarymaterials.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Predictors of smartphone addiction in adolescents with depression: combing the machine learning and moderated mediation model approach

Status:

Version 1

Abstract

Figures

1 Introduction

1.1 Predictors of SA

1.1.1 Predisposing factors

1.1.2 Mediating factors

1.1.3 Moderating factors

1.1.4 Demographics

1.2 Machine learning

1.3 Aim

2 Methods

2.1 Participants

2.2 Measures

2.2.1 Smartphone addiction (SA)

2.2.2 Depression

2.2.3 Anxiety

2.2.4 Sleep problems

2.2.5 Stress perception

2.2.6 Loneliness

2.2.7 Alexithymia

2.2.8 Self-esteem

2.2.9 Rumination

2.2.10 Resilience

2.2.11 Coping

2.2.12 Emotion Regulation Strategies

2.2.13 Social support

2.2.14 Childhood trauma

2.2.15 School bullying

2.3 Statistical and machine learning analysis

3 Results

3.1 Summary Statistics

3.2 Machine learning

3.3 Path analyses

4 Discussion

5 Limitations

6 Conclusion

Declarations

Author Contribution

Acknowledgement

References

Additional Declarations

Supplementary Files

Status:

Version 1