Transformer models with temporal awareness can predict COPD flare-ups from ventilator data alone, enabling faster detection in home settings without waiting for lab results.
This paper develops a transformer-based model to predict acute exacerbations of COPD using only respiratory data from home ventilators, avoiding delays from clinical lab tests. The model uses time-aware attention to track how symptoms change over time, showing better performance than traditional methods for early detection.