Sage Journals: Discover world-class research

Abstract

Suicide and self-harm crises among high school students are significant public health issues. Previous research has often focused on individual factors in suicide and self-harm crises, neglecting the complex interactions between multiple factors. This study, based on the diathesis-stress model, utilized survey data of 12,472 Chinese high school students and employed machine learning methods to construct a decision tree model. It analyzed the most significant negative life events and coping styles in predicting suicide and self-harm crises, explored the impact of these factors on students, and examined sex differences. The classification tree’s built-in contribution function allowed us to obtain the importance of each variable. Results indicated that the model performed well, with the classification tree demonstrating strong predictive accuracy for self-harm and suicide crises among both male and female students. While the impact of negative life events and coping styles on suicide crises showed cross-sex consistency, sex differences were observed for self-harm crises. Among male students, only interpersonal relationships exceeded the 10% threshold in importance, whereas a wider range of events surpassed this threshold for female students. Coping styles played a critical role for both groups, further underscoring their importance in helping students mitigate crises amid negative events. The decision tree model demonstrated high accuracy in identifying students at risk of suicide and self-harm crises. Through the decision tree model, the study identified several key negative life events and coping styles, offering valuable insights for educators to provide more targeted attention and guidance in intervening in suicide and self-harm crises.

Plain language summary

Predicting suicide or self-harm crisis of Chinese students

Study using reports about the relative importance differences of negative life events and coping styles in the classification of self-harm and suicide crisis among Chinese students of different genders. Why was the study conducted? The study aimed to explore the factors contributing to self-harm and suicide crises among Chinese high school students, given the significant psychological challenges they face, especially with societal pressures and college entrance exams. What did the researchers do? The research team created models to classify self-harm and suicide crises in high school students of different genders. They aimed to identify the impact of negative life events and coping styles on these crises. What did the researchers find? With 14,047 participants, over 10% experienced a suicide crisis, and more than 15% faced a self-harm crisis. For males, interpersonal relationships and fantasy and escape were crucial for self-harm, while academic stress and fantasy were vital for suicide crises. Females, on the other hand, prioritized health adaptation, self-regulation for self-harm, and avoidance, interpersonal relationships, and self-regulation for suicide crises. What do the findings mean? The study successfully developed predictive models to identify high school students at risk of self-harm and suicide crises. Detecting and intervening in these cases could potentially reduce the associated risks timely.

Keywords

suicide crisis self-harm crisis negative life events coping style high school students classification tree

Introduction

Adolescent suicide is a serious public health concern worldwide. According to data from the World Health Organization, 720,000 people die by suicide each year, equating to one death every 40 s; suicide is also a lifelong issue and the third leading cause of death globally among people aged 15 to 29 years (World Health Organization, 2024). Meanwhile, self-harm is one of the strongest predictors of suicide mortality during adolescence (Favril et al., 2022; Geulayov et al., 2018). Before individuals engage in actual suicidal or self-harming behaviors, they often experience suicide and self-harm crises (e.g., seriously contemplating suicide or having a strong desire to self-harm). These crises configure a state of psychological imbalance that may result in fatal consequences (Folkman et al., 1986).

High school students have always been a high-risk group for mental crises, and it is usual for individuals to commit suicide and self-harm behaviors during high school stage (Gandhi et al., 2018). For one, high school students face great pressure from college entrance examinations; for another, they are subject to pressures and setbacks from various aspects such as society, family, and schools (Campos et al., 2019). Concomitantly, the prefrontal cortex has not yet matured in high school students. This impairs their ability to regulate themselves and control their behavior completely (Vijayakumar et al., 2014), hindering their ability to effectively cope with negative life events or setbacks. This makes them more prone to fatal crises (Miller & Prinstein, 2019).

In recent years, the relationship of life events with self-harm and suicidal behaviors has been confirmed by an increasing number of research, especially as the recognition and evaluation of life events has advanced. Various studies have shown that recent life events can independently predict adolescent self-harm behaviors (O’Connor et al., 2010). Furthermore, negative life events, such as stress and problems in interpersonal relationships, are the main reasons adolescents engage in self-harm or suicidal behaviors (Claes et al., 2014; Madge et al., 2011). It remains that not all individuals who experience negative life events experience psychological crises. According to the diathesis-stress model, psychological crises result from the combined effects of external pressures and personal psychological qualities (Brodsky et al., 1997). In current research, stress (i.e., an external pressure) mainly refers to life events (Chen et al., 2011), and coping style is an important psychological factor that individuals use to deal with psychologically-stressful events (Compas et al., 2001; Y. J. Zhang et al., 2005). Therefore, a comprehensive predictive model of suicide and self-harm crises among high school students should also consider both the external pressure factors of negative life events and the psychological factor of coping style.

Importantly, the characteristics of the physiological and psychological development of men and women differ, implying that the coping styles and potential adverse effects of negative life events that students of different sexes face may differ significantly. Flannery et al. (2018) found the greatest sex difference in seeking support and emotional expression, with women being more likely to seek support and emotional expression. Compared to men, women are more sensitive to factors that may lead to self-harm behaviors and are, therefore, more susceptible to them, rendering this a vulnerable group for self-harm (B. Y. Wang et al., 2015). Niu et al. (2013) confirmed the moderating effect of sex on the relationship between negative life events and coping styles. Therefore, it is necessary to explore the effects of negative life events and coping styles on suicide and self-harm crises among men and women separately. Exploring this issue can not only help us understand sex differences in the mechanisms of suicide and self-harm crises but also provide direct guidance for developing related sex-specific intervention programs for high school students.

Negative life events and coping styles are multidimensional. For instance, Liu et al. (1997) divided negative life events into six types (related to relationships, academic pressure, punishment, loss, health adaptation, and others), while Fang et al. (2018) classified coping styles into three types (self-regulation, fantasy & avoidance, and seeking help & problem-solving). Therefore, multiple variables should be considered when exploring the complex relationships among life events, coping styles, adolescent suicide, and self-harm.

Traditional regression analysis is inadequate for analyzing nonlinear feature sets in complex models. Moreover, currently, research has focused on single factors (e.g., overall negative life events or certain negative life events), resulting in relatively simplistic models. Machine learning algorithms come to the fore here as effective tools for processing large samples and multidimensional data. Among them, classification and regression tree (CART) algorithms use a nonlinear logic structure of if-then to recursively divide data by finding features with the best classification effect as nodes, having strong statistical power (Gradus et al., 2020; Wei et al., 2022). In addition, decision tree (DT) algorithms can simultaneously account for complex multivariate relationships between each variable and the outcome variable in a model, determining the optimal split point for each variable and constructing optimal node branches in a hierarchical order based on the importance of the influencing factors. DT models are based on CART, used to predict categorical variables, tolerant of variable attributes, have been widely used in constructing classification models for self-harm and suicide among adolescents, and are expected to be applicable to constructing a psychological crisis classification model for high school students (Donnelly et al., 2023; Wei et al., 2022).

In fact, the DT model has been widely used to predict suicide and self-harm crises in Chinese populations. For example, Pan et al. (2023) used a DT model to predict self-harm crises based on demographic characteristics and family and emotional factors, while Xue et al. (2024) validated the effectiveness of this model in predicting self-harm. Wei et al. (2022) demonstrated that DTs can effectively classify suicide and self-harm crises, with accuracy rates of 72% and 75%, respectively, among middle school students. Previous studies have thus fully validated the effectiveness of DT in predicting suicide and self-harm crises.

Accordingly, this study intends to use a DT algorithm to construct DT classification models for suicide and self-harm crises among high school students of different sexes. Furthermore, we explore the classification effectiveness and relative importance weights of various negative life events and coping styles on suicide and self-harm crises.

Methods

Data Source and Data Exclusion

The data used in this study were obtained from the Science Database of People Mental Health of the Population Health Data Archive (PHDA; DOI: 10.12213/11.A001U.202204.281.V1.0), which follows psychological industry standards, expert consensus, guidelines, and regulations, and uses widely-accepted international and domestic standard scales suitable for the Chinese “norm.” Data collection involved validity and reliability evaluations, and was performed following standardized procedures for both offline and online data collection (e.g., organization, training, and management) using surveys and assessments in mental health contexts. In 2021, the PHDA received the CoreTrustSeal global certification for trusted data repositories. The dataset used in this study included high school students from 24 provinces in China, and a total of 14,047 students completed the relevant questionnaire. This database was approved by the ethics committee of cancer hospital, Chinese academy of medical sciences and all the participants provided informed consent, with the latest revision date being May 26, 2023.

Because machine learning models do not allow for missing values, and to improve data quality, we filtered the sample. The exclusion criteria were as follows: (1) questionnaires with missing items; (2) questionnaires with clearly-distorted responses, such as selecting the same option for all questions or selecting answers outside the prescribed range. Based on these criteria, 1,575 (11.21%) students were excluded, leading to12,472 high school students (5,507 men, 44.15%; 6,965 women, 55.85%) being included in the analysis.

In this study, four questions were used to investigate whether high school students had experienced a suicide or self-harm crisis in the past 6 months. The questions included “Seriously considered suicide,”“Formulated a suicide plan,”“Attempted suicide,” and “Engaged in self-harm behavior.” Individuals classified as suicide crisis group (834 men and 1,169 women) reported having suicidal ideation, a suicide plan, or a suicide attempt. Individuals classified as self-harm crisis group (880 men and 1,632 women) reported engaging in self-harm behavior without suicidal ideation, suicide plan, or suicide attempt. The remaining individuals without a suicide or self-harm crisis (3,793 men and 4,164 women) were referred to as the control group.

The dataset included data of 14,047 high school students, from 24 provinces in China, who completed the relevant questionnaire. This research was approved by the Ethics Committee, and all participants provided informed consent, with the latest revision date being May 26, 2023.

Measure

Negative Life Events

Negative life events were measured using the 27-item Chinese version of the Adolescent Self-Rating Life Events Checklist (also known as ASLEC; Liu et al., 1997). Various classification methods have been proposed for the dimensions of this construct. After conducting confirmatory factor analysis, this study adopted the six-dimensional classification criteria proposed by Gao (2006), which includes the following dimensions: interpersonal relationships (five items), academic pressure (five items), punishment (five items), loss (four items), health adaptation (five items), and other (three items; e.g., “financial difficulties in the family”). Items were responded on a five-point Likert scale, with higher scores indicating a greater impact of negative life events. Previous research conducted with the Chinese population has demonstrated this scale to have good reliability and validity (Song et al., 2020; Zou et al., 2022).

Coping Style

Coping style was measured using the 20-item Chinese version of the Simplified Coping Style Questionnaire (SCSQ, Xie, 1998). All items were rated on a four-point Likert scale. The original scale was divided into two dimensions, namely positive and negative coping. However, considering the uniqueness of the psychological characteristics of adolescents, we chose the three-dimensional division standard proposed by Fang et al. (2018), which includes self-regulation (eight items), fantasy and avoidance (seven items), and seeking help and problem-solving (five items; e.g., “Talking to people and sharing inner troubles”). The higher the score, the more frequent the use of the coping style. The SCSQ has shown good reliability and validity in Chinese populations (Li et al., 2022; Lin et al., 2020).

Data Analysis

In this study, SPSS version 22.0 was used for the correlation analysis, F-test, internal consistency reliability analysis, and Harman’s single factor test. Amos version 22.0 was used for confirmatory factor analysis. The CART model was constructed using scikit-learn version 1.1.0 in Python version 3.8.

We used the machine learning CART algorithm to construct classification tree models with negative life events and coping styles as predictor variables, and suicide and self-harm crises as outcome variables.

The model selected nine features, as follows: (1) negative life events (interpersonal relationships, academic stress, punishment, loss, health adaptation, and others); (2) coping styles (self-regulation, fantasy and escape, seeking help, and problem solving). Considering sex differences, we constructed prediction models for self-harm and suicide crises separately for male and female students.

The data were randomly divided into training (70%) and test sets (30%). Because both individuals with suicidal ideation and those with self-harm behaviors belonged to the minority in the dataset, which deviated from the assumption of unbiasedness, oversampling techniques were used to balance the data and generate new samples for the minority classes. The balanced data sizes are listed in Table 1.

Table 1.

Data Size of Suicide Crisis and Self-Harm Crisis.

	Dependent variable	Data set	N
	Dependent variable	Data set	Yes = 1	No = 0	Sum
male	Suicide crisis	Training set	581	3,273	3,854
		Training set after oversampling	2,245	3,286	5,531
		Test set	253	1,400	1,653
	Self-harm crisis	Training set	639	3,215	3,854
		Training set after oversampling	3,211	3,241	6,452
		Test set	241	1,412	1,653
female	Suicide crisis	Training set	808	4,067	4,875
		Training set after oversampling	4,063	4,067	8,130
		Test set	361	1,729	2,090
	Self-harm crisis	Training set	1,130	3,745	4,875
		Training set after oversampling	3,773	3,703	7,476
		Test set	502	1,588	2,090

The model was constructed using the training dataset. In the recursion process, the Gini coefficient was used as a measure of heterogeneity. To reduce overfitting, this study used 10-fold cross-validation to evaluate the model, using evaluation metrics of accuracy, precision, recall, and F1 score. The “feature_importances_” function is used to measure the contribution of each feature to model prediction accuracy, and to assess feature importance. The function returns a value between zero and one, with higher values indicating that the corresponding feature contributes more to the prediction results.

Results

Reliability and Validity Test of the Scales

In this study, the Cronbach’s α coefficients for the Adolescent Self-Rating Life Events Checklist scale and the dimensions (interpersonal relationships, academic pressure, punishment, loss, health adaptation, and other) were .97, .85, .83, .94, .89, .84, and .83, respectively. The Cronbach’s α coefficients for the SCSQ scale and its dimensions (self-regulation, fantasy and avoidance, and seeking help and problem-solving) were .93, .92, .89, and .90, respectively. The scale demonstrated good internal consistency and reliability.

The results of the confirmatory factor analysis are presented in Table 2. For large samples (n ≥ 1,000), the χ² value tends to be large; however, as long as several other indices are satisfactory, the model can be considered acceptable to a certain extent (Wen et al., 2004). Therefore, both the Adolescent Self-Rating Life Events Checklist and SCSQ ideally reflect the constructs of the overall scale.

Table 2.

Results of the Confirmatory Factor Analysis.

Questionnaires	NFI	CFI	GFI	AGFI	TLT	RMSEA
AELEC	0.83	0.83	0.85	0.88	0.81	0.07
SCSQ	0.87	0.88	0.81	0.79	0.86	0.08

Harman’s single-factor test was used to test for common method bias, and the results showed that there were six factors with eigenvalues greater than one. The explanatory rate of the first factor was 29%, which is below the critical value of 40%, indicating no significant common method bias.

Correlation Analysis

The results showed that suicide and self-harm crises correlated significantly with each dimension of negative life events and of coping styles (p < .001), and Figure 1 shows the correlation heatmap. These results of the correlation analysis provided a reliable foundation for model building, as negative life events and coping styles were highly correlated with suicide and self-harm crises, suggesting that the information contained in these variables may have a strong explanatory power in the predictive model.

Figure 1.

Correlation heatmap.

The Results of the Difference Test

The results of χ² and F tests showed significant differences by sex for all dimensions of negative life events and coping styles among the three groups. Post hoc comparisons using the Student-Newman-Keuls method indicated significant differences in all variables among the three groups. The crisis group (suicidal crisis + self-harm crisis) scored higher than the control group for all six negative life events. Regarding coping styles, the crisis group was more inclined toward fantasy and avoidance (Table 3).

Table 3.

The Results of the Difference Test.

Variables	Suicide crisis	Self-harm crisis	Control group	χ²/F	η²
Gender				502.55^***
Male	834	880	3,793
Female	1,169	1,632	4,164
Coping: Self-regulation	−0.058	−0.178	0.010	89.73^***	.014
Coping: Fantasy and avoidance	0.131	0.191	−0.161	189.98^***	.030
Coping: Seeking help and problem solving	−0.090	−0.166	0.098	88.26^***	.014
Negative life events: Interpersonal relationships	0.258	0.367	−0.297	697.62^***	.101
Negative life events: Study pressure	0.273	0.323	−0.289	633.58^***	.092
Negative life events: Punishment	0.193	0.270	−0.323	662.08^***	.096
Negative life events: Loss	0.188	0.265	−0.310	590.92^***	.087
Negative life events: Health adaptation	0.217	0.307	−0.324	715.12^***	.103
Negative life events: Other	0.230	0.293	−0.312	646.53^***	.094

***

p < .001

Model Performance

According to the CART algorithm, the predictive model for whether male students are in a suicide crisis has an accuracy of 81.26%, precision of 78.13%, recall of 86.35%, and an area under the curve (AUC) is 81.30%. Meanwhile, the predictive model for whether male students are in a self-harm crisis has an accuracy of 78.53%, precision of 73.95%, recall of 88.51%, and an AUC of 78.50%.

The predictive model for whether female students are in a suicide crisis has an accuracy of 79.56%, precision of 76.06%, recall of 86.56%, and an AUC of 79.53%. Meanwhile, the predictive model for whether female students are in a self-harm crisis has an accuracy of 77.10%, precision of 73.80%, recall of 84.10%, and an AUC of 77.10%.

Feature Importance

The importance of each feature in the classification tree with self-harm crisis as the outcome variable is shown in Figure 2.

Figure 2.

Feature importance for predicting self-harm crisis.

The results showed that, compared to the model for male students, the DT model identified more features with importance greater than 10% among female students. Among male students, the most important negative life event variable was interpersonal relationships, while the most important coping style variable was fantasy and escape. Among female students, the most important negative life event variable was health adaptation, while the most important coping style variable was self-regulation.

The importance of each feature in the classification tree with suicide crisis as the outcome variable, is shown in Figure 3.

Figure 3.

Feature importance for predicting suicide crisis.

The results showed that the DT model selected five features with an importance greater than 10% in both male and female students. Among male students, the most important negative life event variable was academic stress, while the most important coping variable was fantasy and avoidance. Among female students, the most important negative life event variable was interpersonal relationships, while the most important coping variable was self-regulation.

Discussion

This study constructed complex classification models for suicide and self-harm crises in male and female high school students using a DT algorithm, and explored the relative importance of negative life events and coping style on such crises. The results showed that the DT effectively classified suicide and self-harm crises.

The Effect of Coping Style in Suicide and Self-Harm Crises

According to the feature importance results obtained from the DT model, regardless of a suicide or a self-harm crisis, the contribution of the three coping styles in both male and female students was greater than 10%. This finding suggests that how individuals cope with negative life events is a significant factor in whether they experience suicide or self-harm crises.

Previous studies support this finding. According to the experiential avoidance model, there is a stronger connection between negative life events and deliberate self-harm among adolescents, as they are unable to self-regulate effectively (Chapman et al., 2006). When faced with setbacks, adjusting one’s goals can effectively reduce suicidal ideation, whereas individuals unable to self-regulate when they cannot achieve their goals are more likely to develop suicidal ideation (O’Connor et al., 2010). Additionally, Zou et al. (2022) found that fantasy and escape significantly increase self-harm risks among adolescents, and these findings are also supported by the experiential avoidance model. This model suggests that provocative life events trigger negative emotions and psychological distress, which then lead to an increase in the tendency to avoid psychological experiences. This avoidance can then help individuals avoid psychological distress. To achieve experiential avoidance, individuals may engage in harmful behaviors, such as suicide and self-harm, as a short-term attention diversion (Chapman et al., 2006).

Additionally, seeking help and problem-solving are critical coping styles that serve as protective factors against self-harm behaviors (Wu & Liu, 2019). Interestingly, seeking help and problem-solving (i.e., a positive coping style) was less important than self-regulation in our sample. We speculate that this may be because the commonly-occurring enhanced self-centeredness during adolescence increases high school students’ independence and self-esteem, making them less likely to ask others for help (C. H. Guo, 2019; Z. Wang & Yang, 2005).

The Effect of Negative Life Events in Self-Harm Crises

Regarding negative life events, interpersonal relationships played a critical role in predicting self-harm crises in both male and female students. Still, the role of interpersonal relationship-related negative life events was higher (vs. other negative life events) among male students. One explanation is that interpersonal conflicts can cause individuals to be dissatisfied with their relationships, leading to feelings of loneliness (Lodder et al., 2017; Mahon et al., 2006). Meanwhile, high levels of loneliness can cause individuals to have more negative expectations of social interactions and to be more sensitive to threats (Cacioppo & Hawkley, 2009). In general, male students are more sensitive to negative information in peer interactions and are more prone to conflicts, which leads to further increases in subsequent friend conflicts (S. Y. Hu et al., 2019).

Unlike male students, the negative life events that triggered self-harm crises in female students were more diverse, and this may be related to sex differences in rumination. Studies show that negative life events correlate positively with an individual’s level of rumination (Xin et al., 2023), and that rumination is an important influencing factor in self-harm (Y. L. Wang et al., 2020). Additionally, female students often adopt rumination as a coping style, making the impact of negative life events on female students more pronounced (S. R. Guo et al., 2011; Han & Yang, 2009).

The Role of Negative Life Events in Suicide Crises

The DT identified two features, namely interpersonal relationships and academic pressure, as having an importance greater than 10% in predicting suicide crises in both male and female students, demonstrating consistency across sexes.

Unlike for self-harm crisis prediction, interpersonal relationships were found to be more important in suicide crisis prediction among female than male students, which is a novel finding. While previous research has established a close relationship between interpersonal relationships and psychological crises (Blum et al., 2024), there has been a lack of systematic comparisons examining how interpersonal relationships differently impact suicide crises and self-harm crises across genders. This study addresses this gap by demonstrating that interpersonal relationships influence suicide crises and self-harm crises differently in male and female students, highlighting significant gender-based differences. Previous studies have found that interpersonal conflict has varying impacts on students of different genders. S. Y. Hu et al. (2019) found that while both male and female students may feel lonely owing to interpersonal conflicts, these conflicts mainly manifest through peer conflict and parent–child conflicts did not have a significant predictive influence on loneliness among male students. In contrast, female students’ feelings of loneliness were more likely to be influenced by parent–child conflicts, and there was no significant predictive relationship between peer conflicts and feelings of loneliness (S. Y. Hu et al., 2019). When faced with parent–child conflicts, female students tend to resist and attack more than male students, which further exacerbates the sense of estrangement between parent and child (Y. Hu et al., 2023). As the correlation between parent–child relationships and suicide crises was much higher than that between such relationships and self-harm crises, interpersonal relationship-related negative interpersonal events may be more significant in predicting suicide crises in female students.

In addition, we found that the importance of academic stress in predicting suicide crises in male students was significantly higher than that of other negative life events, which verifies the significance of academic stress in male students. First, academic problems are the main factor associated with male students experiencing family conflicts (S. Y. Hu et al., 2019). Second, academic stress can lead to a decrease in students’ self-esteem and -efficacy (Tao, 2013), and high school male students have a significantly higher need for self-esteem and self-worth than do female students (Shukla & Kang, 2017). A lack of self-worth can directly lead to suicidal ideation (P. Zhang et al., 2022).

This study also expands upon the diathesis-stress model. Specifically, we used a DT model to compare the impact of negative life events and coping styles across sexes. This enabled us to identify the key negative life events that influence the risk of suicide or self-harm crises, and the most effective coping styles for each sex group. Our contribution lies in analyzing large-scale population data, revealing that specific negative life events play different roles across sexes. For example, health adaptation plays a more significant role in self-harm among female students than male students. There are two main explanations for these sex differences: gender socialization theory (i.e., gender roles are shaped by societal and cultural expectations) and biological sex theory (i.e., sex behavior differences caused by hormonal and physiological differences; Hyde, 2005). Future research should explore the reasons for sex differences to further deepen the diathesis-stress model.

This study also highlights the critical importance of coping styles, as they were highly important across all models. This indicates that selecting appropriate coping styles when facing negative life events is crucial for crisis prevention among adolescents. This study deepens our understanding of the relationship between negative life events and coping styles, and provides a theoretical foundation for designing sex-based suicide and self-harm prevention and intervention strategies, offering significant practical value.

Limitations

Despite the strengths of this study, the results should be interpreted in light of the following limitations. First, it was based on cross-sectional data, implying that the predictive model developed for suicide and self-harm crisis thoughts does not infer causality. Further longitudinal studies are required to address the causal relations of these variables.

Second, we translated suicide and self-injury crisis outcomes into a dichotomous variable because this was expected to increase the simplicity and interpretability of the results and further facilitate discussions with a wider range of stakeholders and practitioners. However, some information may have been lost owing to this methodological choice.

Conclusions

Given the seriousness of adolescent suicide and self-harm, prevention and intervention efforts are crucial. Therefore, this study, based on the diathesis-stress model, explored a complex model of negative life events and coping styles for predicting suicide and self-harm crises among high school students of different sexes, and advances our theoretical foundation of the association between self-harm and suicide.

We found that coping style plays a crucial role in predicting both suicide and self-harm crises, highlighting the significance of students’ subjective agency in mitigating these crises. Moreover, we identified important negative life events that contribute to predicting these crises, emphasizing the need for teachers and parents to pay more attention to and provide guidance when students encounter impactful negative life events. Furthermore, among high school students, the predictive model we developed can effectively identify individuals at risk for suicide and self-harm crises; that is, its use may enable timely detection and intervention in this population, potentially reducing the risks associated with suicide and self-harm crises.

Footnotes

Acknowledgements

The data were sourced from the National Center for Medical Informatics Data Warehouse for Population Health and Health Sciences (PHDA), which can be accessed at .

ORCID iD

Shumeng Ma

Ethical Considerations

The study procedures were approved by Hebei Normal University Human Research Ethics Committee.

Author Contributions

Conceptualization: Shu-Meng Ma and Ning Jia; methodology: Shu-Meng Ma and Ning Jia; software: Shu-Meng Ma; formal analysis, Shu-Meng Ma; investigation: Shu-Meng Ma; resources: Ning Jia; data curation, Shu-Meng Ma; writing—original draft preparation: Shu-Meng Ma; writing—review and editing: Ning Jia, Shu-Meng Ma, and Ping Li; visualization: Shu-Meng Ma and Li-wen, Ren; supervision: Ning Jia; project administration: Ning Jia; funding acquisition: Ning Jia All authors have read and agreed to the published version of the manuscript.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the major project of Humanities and Social Sciences of the Hebei Provincial Department of Education (No. ZD202109).

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data that support the findings of this study are available from the Population Health Data Archive (). However, restrictions apply to the availability of these data, which were used under a license for the current study and are not publicly available. The data will be made available from the authors upon reasonable request and with permission from the Population Health Data Archive

References

Blum

Akhavan

Rogers

M. L.

Astudillo-García

C. I.

Cinka

Yilmaz

F. K.

Barzilay

(2024). The role of interpersonal stressors and connectedness in acute suicide risk and the suicide crisis syndrome during the COVID-19 pandemic. Journal of Affective Disorders, 354, 404–425.

Brodsky

B. S.

Malone

K. M.

Ellis

S. P.

Dulit

R. A.

Mann

J. J.

(1997). Characteristics of borderline personality disorder associated with suicidal behavior. American Journal of Psychiatry, 154(12), 1715–1719. https://doi.org/10.1176/ajp.154.12.1715

Cacioppo

J. T.

Hawkley

L. C.

(2009). Perceived social isolation and cognition. Trends in Cognitive Sciences, 13(10), 447–454. https://doi.org/10.1016/j.tics.2009.06.005

Campos

R. C.

Holden

R. R.

Lambert

C. E.

(2019). Avoidance of psychological pain and suicidal ideation in community samples: Replication across two countries and two languages. Journal of Clinical Psychology, 75(12), 2160–2168. https://doi.org/10.1002/jclp.22837

Chapman

A. L.

Gratz

K. L.

Brown

M. Z.

(2006). Solving the puzzle of deliberate self-harm: The experiential avoidance model. Behaviour Research and Therapy, 44(3), 371–394. https://doi.org/10.1016/j.brat.2005.03.005

Chen

Yang

Zhu

X. Z.

Yao

S. Q.

(2011). A diathese-stress model df suicidal ideation in senior high school students: A multiwave longitudinal study. Journal of Psychological Science, 34(03), 733–737. https://doi.org/10.16719/j.cnki.1671-6981.2011.03.029

Claes

Luyckx

Bijttebier

(2014). Non-suicidal self-injury in adolescents: Prevalence and associations with identity formation above and beyond depression. Personality and Individual Differences, 61-62, 101–104. https://doi.org/10.1016/j.paid.2013.12.019

Compas

B. E.

Connor-Smith

J. K.

Saltzman

Thomsen

A. H.

Wadsworth

M. E.

(2001). Coping with stress during childhood and adolescence: Problems, progress, and potential in theory and research. Psychological Bulletin, 127(1), 87–127. https://doi.org/10.1037//0033-2909.127.1.87

Donnelly

H. K.

Han

Kim

Lee

D. H.

(2023). Predictors of suicide ideation among South Korean adolescents: A machine learning approach. Journal of Affective Disorders, 329, 557–565. https://doi.org/10.1016/j.jad.2023.02.079

10.

Fang

Wang

Y. T.

Xiao

S. Y.

Zhao

Jiang

(2018). Reliability and validity of Simplified Coping Style Questionnaire among adolescents. Chinese Journal of Clinical Psychology, 26(5), 905–909. https://doi.org/10.16128/j.cnki.1005-3611.2018.05.015

11.

Favril

Uyar

Sharpe

Fazel

(2022). Risk factors for suicide in adults: Systematic review and meta-analysis of psychological autopsy studies. Evidence-Based Mental Health, 25(4), 148–155.

12.

Flannery

K. M.

Vannucci

Ohannessian

C. M.

(2018). Using time-varying effect modeling to examine age-varying gender differences in coping throughout adolescence and emerging adulthood. Journal of Adolescent Health, 62(3), S27–S34. https://doi.org/10.1016/j.jadohealth.2017.09.027

13.

Folkman

Lazarus

R. S.

Gruen

R. J.

DeLongis

(1986). Appraisal, coping, health status, and psychological symptoms. Journal of Personality and Social Psychology, 50(3), 571–579. https://doi.org/10.1037/0022-3514.50.3.571

14.

Gandhi

Luyckx

Baetens

Kiekens

Sleuwaegen

Berens

Maitra

Claes

(2018). Age of onset of non-suicidal self-injury in Dutch-speaking adolescents and emerging adults: An event history analysis of pooled data. Comprehensive Psychiatry, 80, 170–178. https://doi.org/10.1016/j.comppsych.2017.10.007

15.

Gao

Y. M.

(2006). The confirm factor factor analysis of adolescent self-rating life events check-list among the students in medical vocational colleges. Chinese Journal of Medical Education, 4, 36–39.

16.

Geulayov

Casey

McDonald

K. C.

Foster

Pritchard

Wells

Clements

Kapur

Ness

Waters

Hawton

(2018). Incidence of suicide, hospital-presenting non-fatal self-harm, and community-occurring non-fatal self-harm in adolescents in England (the iceberg model of self-harm): A retrospective study. Lancet Psychiatry, 5(2), 167–174.

17.

Gradus

J. L.

Rosellini

A. J.

Horváth-Puhó

Street

A. E.

Galatzer-Levy

Jiang

Lash

T. L.

Sørensen

H. T.

(2020). Prediction of sex-specific suicide risk using Machine Learning and Single-Payer Health Care Registry data from Denmark. JAMA Psychiatry, 77(1), 25–34. https://doi.org/10.1001/jamapsychiatry.2019.2905

18.

Guo

C. H.

(2019). The relationship between college students’ ego-centrism and subjective well-being: The mediating effect of self-esteem. Studies of Psychology and Behavior, 17(04), 546–552.

19.

Guo

S. R.

X. C.

Guo

Y. Q.

Wang

Tanhg

S. Y.

(2011). The influence of rumination to negative and positive affection: Based on its relationships with loneliness and emotional intelligence. Psychological Development and Education, 27(03), 329–336. https://doi.org/10.16187/j.cnki.issn1001-4918.2011.03.008

20.

Han

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.

21.

S. Y.

Liang

L. S.

Yuan

K. M.

Bian

Y. F.

(2019). The influences of conflicts with parents and friends on loneliness in middle school and its gender difference: A cross-lagged analysis. Journal of Psychological Science, 42(03), 598–603. https://doi.org/10.16719/j.cnki.1671-6981.20190313

22.

Zeng

Peng

Wang

Liu

Yang

Fang

(2023). The effects of the parent-child relationship and parental educational involvement on adolescent depression, self-injury, and suicidal ideation: The roles of defeat and meaning in life. Acta Psychologica Sinica, 55(01), 129–141.

23.

Hyde

J. S.

(2005). The gender similarities hypothesis. American Psychologist, 60(6), 581–592. https://doi.org/10.1037/0003-066X.60.6.581

24.

Fan

Wang

Huang

(2022). Risk factors of psychological distress during the COVID-19 pandemic: The roles of coping style and emotional regulation. Journal of Affective Disorders, 299, 326–334. https://doi.org/10.1016/j.jad.2021.12.026

25.

Lin

Liu

Wang

Wei

Zhu

Chen

Tian

Zhang

Wang

Zhang

Wang

(2020). Perceived stressfulness mediates the effects of subjective social support and negative coping style on suicide risk in Chinese patients with major depressive disorder. Journal of Affective Disorders, 265, 32–38. https://doi.org/10.1016/j.jad.2020.01.026

26.

Liu

X. C.

Oda

Peng

Asai

(1997). Life events and anxiety in Chinese medical students. Social Psychiatry and Psychiatric Epidemiology, 32, 63–67.

27.

Lodder

G. M. A.

Scholte

R. H. J.

Goossens

Verhagen

(2017). Loneliness in early adolescence: Friendship quantity, friendship quality, and dyadic processes. Journal of Clinical Child & Adolescent Psychology, 46(5), 709–720. https://doi.org/10.1080/15374416.2015.1070352

28.

Madge

Hawton

McMahon

E. M.

Corcoran

De Leo

de Wilde

E. J.

Arensman

(2011). Psychological characteristics, stressful life events and deliberate self-harm: findings from the Child & Adolescent Self-harm in Europe (CASE) Study. European Child & Adolescent Psychiatry, 20(10), 499–508. https://doi.org/10.1007/s00787-011-0210-4

29.

Mahon

N. E.

Yarcheski

T. J.

Cannella

B. L.

Hanks

M. M.

(2006). A meta-analytic study of predictors for loneliness during adolescence. Nursing Research, 55(6), 446–447. https://doi.org/10.1097/00006199-200611000-00009

30.

Miller

A. B.

Prinstein

M. J.

(2019). Adolescent suicide as a failure of acute stress-response systems. Annual Review of Clinical Psychology, 15, 425–450.

31.

Niu

G. F.

Hao

E. H.

Sun

X. J.

Zhou

Z. K.

(2013). Negative life events’ impact on depression among college students: The mediating effect of coping and the moderating effect of gender. Chinese Journal of Clinical Psychology, 21(6), 1022–1025. https://doi.org/10.16128/j.cnki.1005-3611.2013.06.016

32.

O’Connor

R. C.

Rasmussen

Hawton

. (2010). Predicting depression, anxiety and self-harm in adolescents: The role of perfectionism and acute life stress. Behaviour Research and Therapy, 48(1), 52–59. https://doi.org/10.1016/j.brat.2009.09.008

33.

Pan

Liu

Shi

Zhao

Tian

Chen

Zhang

(2023). A machine learning-based predictive model of non-suicidal self-injurious behavior among college students in Guizhou Province. Chinese Journal of School Health, 8, 1198–1202.

34.

Shukla

Kang

T. K.

(2017). Self-esteem among adolescents: A study of gender and locale differences. Indian Journal of Health and Wellbeing, 8(9), 981–984.

35.

Song

Pan

Tao

Ren

(2020). Associations between screen time, negative life events, and emotional and behavioral problems among Chinese children and adolescents. Journal of Affective Disorders, 264, 506–512. https://doi.org/10.1016/j.jad.2019.11.082

36.

Tao

(2013). Self-efficacy in relation to mental health among senior high school students. Chinese Journal of School Health, 34(11), 1333–1335. https://doi.org/10.16835/j.cnki.1000-9817.2013.11.020

37.

Vijayakumar

Whittle

Dennison

Yücel

Simmons

Allen

N. B.

(2014). Development of temperamental effortful control mediates the relationship between maturation of the prefrontal cortex and psychopathology during adolescence: A 4-year longitudinal study. Developmental Cognitive Neuroscience, 9, 30–43. https://doi.org/10.1016/j.dcn.2013.12.002

38.

Wang

B. Y.

Zhang

M. Q.

Zhang

J. T.

You

J. N.

Liang

Y. J.

(2015). Latent transition analysis of the self-injury in adolescents: A longitudinal study. Journal of Psychological Science, 38(06), 1368–1376. https://doi.org/10.16719/j.cnki.1671-6981.20150613

39.

Wang

Y. L.

Zhang

M. Q.

Tan

G. L.

Lin

(2020). The relationship between cybervictimization and selfinjury of adolescents: The moderating role of friendship quality and ruminative response. Journal of Psychological Science, 43(2), 363–370. https://doi.org/10.16719/j.cnki.1671-6981.20200215

40.

Wang

Yang

J. X.

(2005). Coping style and mental health of senior middle school students. Chinese Journal of School Health, 26(10), 824–825.

41.

Wei

S. J.

H. H.

Sun

(2022). Psychological crisis in Chinese adolescents: Using a classiﬁcation tree approach to combine the three dimensional psychological pain model. Journal of Psychological Science, 45(3), 732–739. https://doi.org/10.16719/j.cnki.1671-6981.20220330

42.

Wen

Hau

K. T.

Herbert

W. M.

(2004). Structural equation model testing: Cutoff criteria for goodness of fit indices and Chi-square test. Acta Psychologica Sinica, 36(2), 186–194.

43.

World Health Organization. (2024). Suicide prevention. World Health Organization. https://www.who.int/health-topics/suicide#tab=tab_1

44.

Liu

(2019). Features of nonsuicidal self-injury and relationships with coping methods among college students. Iranian Journal of Public Health, 48(2), 270–277.

45.

Xie

Y. N.

(1998). Reliability and validity of the simplified Coping Style Questionnaire. Chinese Journal of Clinical Psychology, 6(2), 114–115.

46.

Xin

S. F.

Wang

J. R.

Peng

H. Y.

L. Q.

Zhang

Y. L.

Sheng

(2023). The influence of negative life events on adolescent anxiety: The perspective of cognitive behavioral therapy. Psychological Development and Education, 39(5), 710–717. https://doi.org/10.16187/j.cnki.issn1001-4918.2023.05.12

47.

Xue

Ziyuan

Jing

Hui

(2024). Classification decision tree analysis on influencing factors of suicidal behavior in college students. Psychological Development and Education, 40(3), 421–430.

48.

Zhang

Wang

Zhang

L. G.

(2022). Effect of self-concept clarity and meaning in life on suicidal ideation in college fresh students. Chinese Mental Health Journal, 36(11), 975–980.

49.

Zhang

Y. J.

Yan

K. L.

Wang

J. L.

(2005). A path analysis on life events, negative automatic thoughts, coping style and depression. Psychological Development and Education, 21(1), 96–99.

50.

Zou

Song

Tan

Deng

Zhang

Tao

Tang

Xiong

Huang

Yang

Zeng

Liu

Shen

Zhao

Chen

… Yin

(2022). Core self-evaluation as mediator between depressive symptoms and suicidal ideation in adolescents. Journal of Affective Disorders, 302, 361–366. https://doi.org/10.1016/j.jad.2022.01.093

Predicting Suicide or Self-Harm Crises Based on Decision Tree Analysis of Life Events and Coping Style: A Population-Based Study in China

Abstract

Plain language summary

Keywords

Introduction

Methods

Data Source and Data Exclusion

Measure

Negative Life Events

Coping Style

Data Analysis

Results

Reliability and Validity Test of the Scales

Correlation Analysis

The Results of the Difference Test

Model Performance

Feature Importance

Discussion

The Effect of Coping Style in Suicide and Self-Harm Crises

The Effect of Negative Life Events in Self-Harm Crises

The Role of Negative Life Events in Suicide Crises

Limitations

Conclusions

Footnotes

Acknowledgements

ORCID iD

Ethical Considerations

Author Contributions

Funding

Declaration of Conflicting Interests

Data Availability Statement

References