Leveraging Periodicity for Robustness with Multi-modal Mood Pattern Models

Data from wearable sensors (e.g., heart rate, step count) can be used to model mood patterns. We characterize feature representations and modeling strategies with multi-modal discrete time series data for mood pattern classification with a large dataset with naturalistic missingness (n=116,819 participants) using 12 wearable data streams, with a focus on capturing periodic trends in data. Considering both performance and robustness, periodicity-based aggregate feature representations with gradient boosting models outperformed other representations and architectures studied. The use of periodic features improved the model performance compared to temporal statistics, and gradient boosting models were more robust to dataset and shifts in missingness distributions than a deep learning time series model.