Are false positives in suicide classification models a risk group? Evidence for "true alarms" in a population-representative longitudinal study of Norwegian adolescents

Differentiating adolescent suicidal and nonsuicidal self-harm with artificial intelligence: Beyond suicidal intent and capability for suicide

Clinical differentiation between adolescent suicidal self-harm (SSH) and nonsuicidal self-harm (NSSH) is a significant challenge for mental health professionals, and its feasibility is controversial. The aim of the present study was to determine whether adolescent SSH can be differentiated from NSSH based on a holistic multimodel domain analysis (HMDA) of individual, psychological, sociodemographic, and environmental data without relying on intent and acquired capability for suicide. The study also sought to compare the compatibility of the HMDA result with four theories of suicide: the interpersonal theory of suicide, the three-steps theory, the integrated motivational-volitional model, and the theory of adolescent suicidality and self-destruction. This cross-sectional study analyzed data from a nationwide survey conducted from 2017 to 2019 in Norway. The sample included 26,904 adolescents (mean age = 15.01; 29.9 % males), who self-reported suicidal and nonsuicidal self-harm. The meta-learner classification model reached an AUC of 77.0 % and a 70.0 % for both sensitivity and specificity in differentiating the two conditions. Exploratory factor analysis suggested eight factors for clustering the important risk and protective indicators: 1) disordered eating and future orientation, 2) sleep disturbance, 3) orientation toward affects and self, 4) risk-taking behavior, 5) peer relationships, 6) victimization, safety, and security, 7) family relationships, and 8) substance use. These factors emphasize that domains central to adolescent physical, personal, social, and expected future well-being provide invaluable information about risk of suicidal and self-destructive behavior. The results also indicated that despite their high comorbidity, adolescent SSH and NSSH are multifaceted and can be differentiated without relying on suicide-related information. These findings fully supported the theory of adolescent suicidality and self-destruction and partially supported the ideation-to-action theories.

Risk prediction models for adolescent suicide: A systematic review and meta-analysis

BACKGROUND

Adolescence is recognized as a high-risk period for suicide, with the prevalence of suicide risk among adolescents rising globally, positioning it as one of the most urgent public health concerns worldwide. This study systematically reviews and evaluates adolescent suicide risk prediction models, identifies key predictors, and offers valuable insights for the development of future tools to assess suicide risk in adolescents.

METHODS

We systematically searched four international databases (PubMed, Web of Science, Embase, and Cochrane Libraries) and four Chinese databases (Chinese Biomedical Literature Database, China National Knowledge Infrastructure, Wanfang, and Weipu Libraries) up to May 14, 2024. Two researchers independently screened the literature, extracted data, and evaluated the model quality using the Prediction Model Risk of Bias Assessment Tool (PROBAST). Stata17.0 and R4.4.2 softwares were used to conduct meta-analysis.

RESULTS

25 studies involving 62 prediction models were included, of which 51 models were internally validated with an area under the curve (AUC) > 0.7. The researchers mainly used modeling methods such as logistic regression (LR), random forest (RF), extreme gradient boosting (XGBoost), decision tree (DT), and support vector machine (SVM). 22 studies performed internal validation of the model, while only 3 had undergone external validation. The models developed in all 25 studies demonstrated good applicability, 19 studies showed a high risk of bias, primarily due to inappropriate data sources and poor reporting of the analysis domain. Meta-analysis results showed that the pooled AUC for internal validation of 28 adolescent suicide risk prediction models was 0.846 (95 %CI=0.828-0.866), while the AUC for external validation of 2 models was 0.810 (95 %CI=0.704-0.932). The detection rate of suicide risk among adolescents was 22.5 % (95 %CI=18.0 %-27.0 %), gender(OR=1.490,95 %CI=1.217-1.824), depressive symptoms (OR=3.175,95 %CI=1.697-5.940), stress level (OR=2.436,95 %CI=1.019-5.819), previous suicidal ideation (OR=1.772,95 %CI=1.640-1.915), previous self-injurious behaviors (OR=4.138,95 %CI=1.328-12.895), drug abuse(OR=3.316,95 %CI=1.537-7.154), history of bullying(OR=3.417,95 %CI=2.567-4.547), and family relationships (OR=1.782,95 %CI=1.115-2.849) were independent influences on adolescent suicide risk (P < 0.05).

CONCLUSION

The adolescent suicide risk prediction model demonstrated excellent predictive performance. However, given the high risk of bias in most studies and the insufficient external validation, its clinical applicability requires further investigation. Future studies on adolescent suicide risk prediction models should focus on predictors, including gender, depressive symptoms, stress level, previous suicidal ideation, previous self-injurious behaviors, drug abuse, history of bullying, and family relationships.

Suicide attempt risk predicts inconsistent self-reported suicide attempts: A machine learning approach using longitudinal data

INTRODUCTION

Inconsistent self-reports of lifetime suicide attempts (LSAs) are a major obstacle for accurate assessment of suicidal behavior. This study is the first to posit that adolescents at higher risk report LSAs more consistently than those at lower risk, revealing a link between suicide attempt risk and consistent reporting.

METHODS

A machine learning model was trained with 70 % of the baseline assessment data of a longitudinal sample of Norwegian adolescents (n = 10,739). The model was used to estimate the LSA risk score for the remaining 30 % of the testing dataset. The relationship between these baseline risk scores and the consistency of reporting LSAs was assessed using a 2-year follow-up reassessment of the testing dataset.

RESULTS

Internalizing problems, optimism about the future, conduct problems, substance use, and disordered eating were important factors associated with suicide attempt risk. Of the participants, 63.41 % had inconsistent self-reports at the two-year follow-up. Adolescents who consistently reported LSAs had significantly higher scores of suicide attempt risk at baseline. Two logistic regression analyses confirmed an association between suicide attempt risk and inconsistent self-reported LSAs and showed that sex (being male), and lower levels of depression and conduct problems significantly predicted such inconsistencies. Those who inconsistently reported LSAs were more likely than the others to be classified by the model as false negatives at the baseline risk assessment due to their lower estimated risk scores.

LIMITATIONS

Suicide attempts were measured with a single item in this study.

CONCLUSION

These risk factors support the theory of adolescent suicidality (TAS) and could improve suicide attempt risk assessment. Inconsistent self-reported LSAs signal lower suicide attempt risk.

Assessment of Suicide Risk in Patients with Depressive Episodes Due to Affective Disorders and Borderline Personality Disorder: A Pilot Comparative Study

This study assessed suicidal risk in patients suffering from non-psychotic depressive disorders within various clinical and nosological forms (F31-F34 mood disorders and F60.31-emotionally unstable personality disorder). Clinical and psychological features were presented, as well as predictors of suicidal risk in patients of these groups. We performed a comparative analysis of the anxiety and depression level, the level of mental pain, fear of death and the severity of anti-suicidal motives in patients with affective disorders and borderline personality disorder (BPD). Based on the results, 100% of patients in these clinical nosological groups were found to have a high level of suicidal risk. Patients with affective disorders have weak anti-suicidal motives and are not fully aware of the consequences of their own death. Patients with BPD have a higher suicidal risk than patients with affective disorders; they are characterized by less pronounced social orientation, demonstrativeness, self-centeredness, less pronounced levels of anxiety and fear of death.

Unveiling Adolescent Suicidality: Holistic Analysis of Protective and Risk Factors Using Multiple Machine Learning Algorithms

Adolescent suicide attempts are on the rise, presenting a significant public health concern. Recent research aimed at improving risk assessment for adolescent suicide attempts has turned to machine learning. But no studies to date have examined the performance of stacked ensemble algorithms, which are more suitable for low-prevalence conditions. The existing machine learning-based research also lacks population-representative samples, overlooks protective factors and their interplay with risk factors, and neglects established theories on suicidal behavior in favor of purely algorithmic risk estimation. The present study overcomes these shortcomings by comparing the performance of a stacked ensemble algorithm with a diverse set of algorithms, performing a holistic item analysis to identify both risk and protective factors on a comprehensive data, and addressing the compatibility of these factors with two competing theories of suicide, namely, The Interpersonal Theory of Suicide and The Strain Theory of Suicide. A population-representative dataset of 173,664 Norwegian adolescents aged 13 to 18 years (mean = 15.14, SD = 1.58, 50.5% female) with a 4.65% rate of reported suicide attempt during the past 12 months was analyzed. Five machine learning algorithms were trained for suicide attempt risk assessment. The stacked ensemble model significantly outperformed other algorithms, achieving equal sensitivity and a specificity of 90.1%, AUC of 96.4%, and AUCPR of 67.5%. All algorithms found recent self-harm to be the most important indicator of adolescent suicide attempt. Exploratory factor analysis suggested five additional risk domains, which we labeled internalizing problems, sleep disturbance, disordered eating, lack of optimism regarding future education and career, and victimization. The identified factors provided stronger support for The Interpersonal Theory of Suicide than for The Strain Theory of Suicide. An enhancement to The Interpersonal Theory based on the risk and protective factors identified by holistic item analysis is presented.

Predicting suicide attempts among Norwegian adolescents without using suicide-related items: a machine learning approach

Introduction

Research on the classification models of suicide attempts has predominantly depended on the collection of sensitive data related to suicide. Gathering this type of information at the population level can be challenging, especially when it pertains to adolescents. We addressed two main objectives: (1) the feasibility of classifying adolescents at high risk of attempting suicide without relying on specific suicide-related survey items such as history of suicide attempts, suicide plan, or suicide ideation, and (2) identifying the most important predictors of suicide attempts among adolescents.

Methods

Nationwide survey data from 173,664 Norwegian adolescents (ages 13-18) were utilized to train a binary classification model, using 169 questionnaire items. The Extreme Gradient Boosting (XGBoost) algorithm was fine-tuned to classify adolescent suicide attempts, and the most important predictors were identified.

Results

XGBoost achieved a sensitivity of 77% with a specificity of 90%, and an AUC of 92.1% and an AUPRC of 47.1%. A coherent set of predictors in the domains of internalizing problems, substance use, interpersonal relationships, and victimization were pinpointed as the most important items related to recent suicide attempts.

Conclusion

This study underscores the potential of machine learning for screening adolescent suicide attempts on a population scale without requiring sensitive suicide-related survey items. Future research investigating the etiology of suicidal behavior may direct particular attention to internalizing problems, interpersonal relationships, victimization, and substance use.