AttentiveGRUAE: An Attention-Based GRU Autoencoder for Temporal Clustering and Behavioral Characterization of Depression from Wearable Data

The study introduces AttentiveGRUAE, an attention-based gated recurrent unit (GRU) autoencoder aimed at temporal clustering and predicting depression outcomes from wearable data. The model optimizes three objectives: learning a compact latent representation of daily behaviors, predicting end-of-period depression rates, and identifying behavioral subtypes through Gaussian Mixture Model (GMM) clustering. Evaluated on longitudinal sleep data from 372 participants, AttentiveGRUAE outperformed baseline models in clustering quality and depression classification metrics.