Subjective well-being of children with special educational needs: Longitudinal predictors using machine learning

Area of Science:

• Child Psychology
• Educational Psychology
• Machine Learning Applications

Background:

• Children with special educational needs (SEN) exhibit diverse well-being and mental health challenges.
• Understanding well-being predictors requires complex, multifactorial approaches.
• Longitudinal prediction models are crucial for this population.

Purpose of the Study:

• To predict subjective well-being in children with SEN using machine learning.
• To identify key predictors of well-being across diverse SEN profiles.
• To explore complex interactions between predictors and well-being.

Main Methods:

• Longitudinal study of 499 children with diverse SEN (mean age 8.4 years).
• Utilized 32 predictor variables (demographics, life experiences).
• Employed nonlinear machine learning and classical linear classifiers for prediction.

Main Results:

• Nonlinear machine learning significantly outperformed linear classifiers (F1 score 0.72-0.84).
• Key predictors included prior well-being, numeracy, literacy, and interpersonal skills.
• Four distinct clusters emerged, highlighting varied predictor importance (e.g., 'socializer', 'analyzer').

Conclusions:

• Machine learning reveals multiple pathways to well-being in children with SEN.
• Findings inform tailored interventions for supporting SEN well-being.
• Interpersonal and academic factors are critical, but their influence varies by individual profile.