TILES-2018
Citation: Mundnich, K., et al. (2020). Scientific Data, 7, 414.
DOI: 10.1038/s41597-020-00655-3
Access: Download from Scientific Data repository
Tier: 1
Overview
Multi-sensor longitudinal dataset from 212 participants. Designed for tracking behavioral states, emotional well-being, and job performance over time. Combines wearable physiological data with rich survey/self-report data.
Signals
- Empatica E4 wristband (BVP, EDA, ACC, TEMP)
- Smartphone-based surveys (ecological momentary assessment)
- Speech audio from phone
- Activity tracking via phone sensors
Labels
- Self-reported emotional states
- Behavioral states
- Job performance metrics
- Well-being surveys
Relevance
Largest in-the-wild E4 dataset available. Surveys provide weak labels for emotional/arousal states. Multi-week per participant. Useful for training robust models with real-world noise profiles.