MACHINE LEARNING BASED PREDICTION OF DEVELOPMENTAL AND BEHAVIORAL RISK IN EARLY CHILDHOOD: THE ROLE OF SCREEN TIME AMONG MULTIMODAL RISK FACTORS

Abstract

Excessive and early exposure to mobile phones and other digital screens has become a near-universal feature of childhood, coinciding with growing clinical and public-health interest in its relationship to developmental and behavioral outcomes, including autism spectrum disorder (ASD)-related risk indicators. Machine learning (ML) offers a scalable, data-driven approach for identifying children at elevated developmental risk by integrating heterogeneous predictors  screen-time patterns, behavioral questionnaires, eye-tracking metrics, gait and motor indices, voice/ speech features and sociodemographic covariates into unified predictive models. Building on a synthesis of recent literature in this area, we implemented and compared three supervised classifiers Support Vector Machine (SVM), Decision Tree and Random Forest for predicting developmental/autism-related risk from a feature set that included early screen-time exposure alongside behavioral and screening-derived indicators. The Random Forest model achieved the strongest performance, with training accuracy of 96.5% (sensitivity 95.2%, specificity 94.26%, F1-score 0.93) and testing accuracy of 98.10% (sensitivity 97.25%, specificity 95.0%, F1-score 0.95) and corresponding ROC-AUC values above 0.98 on both partitions. We report confusion matrices and ROC curves for the training and testing partitions and discuss how screen-time features can be incorporated responsibly into multimodal risk-stratification pipelines without implying that mobile phone exposure is a causal determinant of autism. We conclude with evaluation criteria, ethical safeguards, preventive/intervention implications, limitations  including the need for external, prospective validation — and directions for future work. The framework and results are intended to support early identification and referral for developmental screening, and to inform evidence-based guidance on screen-time exposure in early childhood.