Temporal Modeling of Cognitive Decline: Health Patterns and Disease Progression in the All of Us Dataset

This study demonstrates the feasibility of deriving cognitive trajectories from All of Us data and reveals comorbidity profiles linked to progression speed. Findings support the development of explainable, fairness-aware AI models to improve early risk prediction. Future work will integrate these models into clinical workflows to support timely, targeted dementia care, particularly for groups historically affected by diagnostic delays and limited access.