E1.S6

DATASET DESCRIPTION

Disease Area

Adolescent Behavioral Health / Clinical Psychiatry

Specific Research Focus

Social withdrawal classification and Major Depressive Disorder (MDD) risk prediction in adolescents aged 12–19

Dataset Theme / Edition

Adolescent Behavioral Health Dataset — ABHD-2026-v1.0

Short Dataset Summary

A fully synthetic, clinically realistic dataset of 1,000 adolescent patient records spanning cross-sectional behavioral profiles and longitudinal therapy session trajectories, generated with evidence-based psychiatric correlations, DSM-5 clinical constraints, and six validated trajectory archetypes

Intended Research Applications

Social withdrawal severity classification, MDD early detection from behavioral signals, CBT response prediction, bullying impact trajectory modeling, school-based intervention timing, relapse risk forecasting, XAI/SHAP behavioral attribution, and federated learning across school and clinic networks

Total Number of Patients

1,000 synthetic patients; 12,912 longitudinal session rows

Number of Features / Parameters

37 cross-sectional features (Part A) + 29 dynamic session-level variables (Part B) = 66 total; NumPy tensor shape (1000 × 18 × 26)

Data Modalities Included

Tabular cross-sectional, longitudinal time-series, ordinal psychometric scales (PHQ-A, SPIN-equivalent, PSQI-equivalent, PSS-10), binary clinical flags, categorical behavioral patterns, and encoded ML-ready tensors with attention masks

Recommended AI Tasks

Multi-class classification, regression, survival analysis, time-series forecasting, sequence modeling (LSTM/Transformer), generative modeling (TimeGAN), explainability (SHAP), and federated learning

Dataset Size

~4.9 MB total across CSV, NumPy (.npy), and metadata files

Release Date

May 2026

Why This Dataset Exists

No publicly available synthetic dataset combines clinically realistic adolescent behavioral health profiles with longitudinal therapy trajectories, DSM-5-enforced severity labels, and ML-ready multi-format exports — leaving researchers without a privacy-safe benchmark for adolescent mental health AI development

Real-World Problem Addressed

Adolescent depression and social withdrawal affect 15–25% of young people globally, yet AI models for early detection and intervention timing remain underdeveloped due to data scarcity, patient privacy constraints, and lack of longitudinal clinical benchmarks — this dataset directly enables research to close that gap