Key Features
A multimodal, longitudinal dataset built to support rigorous evaluation of affective computing models across diverse generalization scenarios.
Sensing Modalities
Multi-year / Multi-wave Collection
Data collected across multiple longitudinal waves spanning extended periods, enabling temporal generalization research and seasonal effect analysis.
Passive Smartphone Sensing
Continuous passive data streams including GPS, accelerometer, screen usage, app logs, and communication metadata — collected without active user effort.
Wearable Sensor Data
Physiological signals from commercial wearable devices, including heart rate, activity counts, sleep metrics, and additional biometric streams.
Labeling & Analysis
Dense Self-Reported Labels
High-frequency ecological momentary assessment (EMA/ESM) captures affective state labels across valence, arousal, stress, and psychological disturbance dimensions.
Longitudinal Analysis Support
Structured to enable intra- and inter-individual longitudinal modeling, including idiographic and nomothetic model evaluation paradigms.
Generalization Benchmarking
Supports four standardized benchmark evaluation tasks — temporal, cross-user, cross-wave, and cross-dataset generalization of learned models.