Longitudinal prediction of youth depression using multivariable machine learning

Depression often emerges during adolescence and is a leading cause of disability. Accurate prediction could inform targeted prevention strategies. This exploratory study examined the predictive accuracy of elastic net regression, identifying key predictors of adolescent depression. Data were drawn from Waves 1-7 (child age 4/5 to 16/17) of the 2004 Longitudinal Study of Australian Children. A total of 174 child- and parent-reported predictors across various domains were used to predict probable depression at age 16–17 years. Elastic net regression with nested cross-validation was implemented. Among 4983 participants, the model achieved an AUC of 0.74. Top predictors included adolescent self-harm, suicidal ideation and plans, lack of secure housing, sex at birth, lack of social support, emotional problems, ADHD diagnosis, and low paternal closeness. Other variables highlighted in previous research, such as parent-child relationship quality, parental warmth, or overall time spent on social media in early adolescence did not emerge among the top predictors. Findings should be viewed in light of several limitations (uncertain generalisability, potential bias from non-independent data, imputation constraints) with replication needed to assess external validity of results. Machine learning demonstrated acceptable predictive ability, highlighting the diverse array of individual, familial, and social risk factors for adolescent depression.

Authors

Institutions

Publication Details

Journal
npj Mental Health Research
Published
2026-11-17
DOI
https://doi.org/10.1038/s44184-026-00257-8
Primary Topic
Child and Adolescent Psychosocial and Emotional Development
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Longitudinal prediction of youth depression using multivariable machine learning

Rachel Visontay, Nicola C. Newton, Lucinda Rachel Grummitt, Tim N Slade et al.
2 links
npj Mental Health Research
Child and Adolescent Psychosocial and Emotional Development
article

Longitudinal prediction of youth depression using multivariable machine learning

Rachel Visontay, Nicola C. Newton, Lucinda Rachel Grummitt, Tim N Slade, Louise Birrell, Toni Coote, Philip J. Clare
article en

Abstract

Depression often emerges during adolescence and is a leading cause of disability. Accurate prediction could inform targeted prevention strategies. This exploratory study examined the predictive accuracy of elastic net regression, identifying key predictors of adolescent depression. Data were drawn from Waves 1-7 (child age 4/5 to 16/17) of the 2004 Longitudinal Study of Australian Children. A total of 174 child- and parent-reported predictors across various domains were used to predict probable depression at age 16–17 years. Elastic net regression with nested cross-validation was implemented. Among 4983 participants, the model achieved an AUC of 0.74. Top predictors included adolescent self-harm, suicidal ideation and plans, lack of secure housing, sex at birth, lack of social support, emotional problems, ADHD diagnosis, and low paternal closeness. Other variables highlighted in previous research, such as parent-child relationship quality, parental warmth, or overall time spent on social media in early adolescence did not emerge among the top predictors. Findings should be viewed in light of several limitations (uncertain generalisability, potential bias from non-independent data, imputation constraints) with replication needed to assess external validity of results. Machine learning demonstrated acceptable predictive ability, highlighting the diverse array of individual, familial, and social risk factors for adolescent depression.

npj Mental Health Research
The University of Sydney (AU), UNSW Sydney (AU), National Drug and Alcohol Research Centre (AU)
NSW Ministry of Health, National Health and Medical Research Council
Good health and well-being
Child and Adolescent Psychosocial and Emotional Development
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon