Predicting Adolescent Mental Health Outcomes Across Cultures: A Machine Learning Approach

Abstract

Adolescent mental health problems are rising rapidly around the world. To combat this rise, clinicians and policymakers need to know which risk factors matter most in predicting poor adolescent mental health. Theory-driven research has identified numerous risk factors that predict adolescent mental health problems but has difficulty distilling and replicating these findings. Data-driven machine learning methods can distill risk factors and replicate findings but have difficulty interpreting findings because these methods are atheoretical. This study demonstrates how data- and theory-driven methods can be integrated to identify the most important preadolescent risk factors in predicting adolescent mental health. Machine learning models examined which of 79 variables assessed at age 10 were the most important predictors of adolescent mental health at ages 13 and 17. These models were examined in a sample of 1176 families with adolescents from nine nations. Machine learning models accurately classified 78% of adolescents who were above-median in age 13 internalizing behavior, 77.3% who were above-median in age 13 externalizing behavior, 73.2% who were above-median in age 17 externalizing behavior, and 60.6% who were above-median in age 17 internalizing behavior. Age 10 measures of youth externalizing and internalizing behavior were the most important predictors of age 13 and 17 externalizing/internalizing behavior, followed by family context variables, parenting behaviors, individual child characteristics, and finally neighborhood and cultural variables. The combination of theoretical and machine-learning models strengthens both approaches and accurately predicts which adolescents demonstrate above average mental health difficulties in approximately 7 of 10 adolescents 3–7 years after the data used in machine learning models were collected.

Brain network connectivity during peer evaluation in adolescent females: Associations with age, pubertal hormones, timing, and status

“…For years, researchers have sought to clarify how adolescent development provides a sensitive window of opportunity for changes to the brain, social behavior, and mental health. Various accounts emphasize pubertal indices such as pubertal timing ( Barendse et al, 2022 , Pfeifer and Allen, 2021b , Ullsperger and Nikolas, 2017 ), whereas others focus on age and the environments associated with being an adolescent ( Costello et al, 2011 , Ortuño-Sierra et al, 2021 , Rothenberg et al, 2023 ). Given that peer evaluation is a particularly salient and shifting experience during puberty, we used a peer evaluation context to probe specificity in how different pubertal indices and age relate to brain connectivity.…”

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

“…Indeed, meta-analyses investigating multiple risk factors for suicide attempts have revealed weak associations, casting doubt on the utility of risk factors studied in isolation with respect to actually predicting and preventing suicide attempts (Bentley et al, 2016 ; Ribeiro et al, 2016 , 2018 ). Failing to consider the interconnectedness of risk and protective factors is likely to result in oversimplified understandings of their mechanisms and to misdirect research and clinical practice (Rothenberg et al, 2023 ). In sum, there is an urgent need for rigorous holistic analysis to better understand both risk and protective factors of suicidal behavior.…”

“…For instance, stacked ensemble algorithms that are particularly effective for addressing low prevalence outcomes such as suicide attempts (see Methods section) have not been utilized for suicide attempt classification (Haghish et al, 2023 ). Furthermore, the separation between ‘theory-driven’ and ‘data-driven’ in mental health studies (Rothenberg et al, 2023 ) is evident among recent machine learning studies on suicidal behavior, which primarily focus on algorithmic risk estimation and overlook established theories, which in turn limits their implications in both research and practice (see for example, Jung et al, 2019 ). Nevertheless, machine learning can make an important contribution to research on suicide risk, insofar as it can be used to identify and rank risk and protective factors in a more comprehensive manner while simultaneously taking a multitude of variables into account (König et al, 2021 ).…”