Romani 2026 — Using Wearable Technology to Predict Severe Behavior Problems: A Systematic Review
Citation: Romani, P.W., D'Mello, S.K., Moulder, R.M., et al. (2026). Perspectives on Behavior Science, 49(2), 361-383.
DOI: 10.1007/s40614-026-00497-1
Type: Systematic review
Key Contribution
Systematic review of 13 peer-reviewed studies on predicting severe behavior problems (SBPs) in neurodevelopmental disabilities using wearables. Highlights methodological concerns and proposes research agenda.
Key Findings
- Most studies claim prediction 30s+ in advance
- Methodological concerns reduce veracity of some claims
- Past behavior often the best predictor of future behavior
- Augmented feature vectors (including contextual aggression info) consistently outperformed physiology-only models
Recommendations
- Open data practices critical for reproducibility
- Social validity data should be collected alongside biometrics
- Need for larger, more diverse populations
Linked To
- Paper: ferina-2025-review
- All Tier 1 datasets and papers