REACT — Real-time Early Agitation Capture Technology (Vanderbilt)
Citation: Khan, N., Plunk, A., Zheng, Z., Adiani, D., Staubitz, J., et al. (2024). Digital Health, 10, 20552076241287884.
DOI: 10.1177/20552076241287884
Access: Contact Vanderbilt research group (Khan/Sarkar) — dataset and pre-trained models available to collaborators
Tier: 1
Overview
The closest existing system to Meltdown Minder. Real-time wearable agitation detection for children with intellectual and developmental disabilities (IDD), including ASD. N=3 children (6-9 yrs), naturalistic settings.
Signals
- HR (heart rate)
- EDA (electrodermal activity)
- ACC (accelerometry)
- Device: unspecified wrist-worn wearable
Performance
- F1 = 78.69% at time of agitation
- F1 = 68.20% 20 seconds prior to agitation
Key Relevance
- Same target population (IDD/ASD children)
- Same detection paradigm (real-time physiological → pre-escalation alert)
- Multimodal sensor fusion (HR + EDA + accelerometry)
- Naturalistic deployment (real-world, not lab)
- Pre-trained models available for transfer learning
Linked To
- Paper: khan-2024-react
- Full study notes