This webinar discusses pros, cons and challenges of moving health informatics research from silos and registry-centric to multi-centric electronic health records data. It will provide tools and data science methods incorporated to mine electronic health records to extract and capture time-dependent features when equally-spaced time is not present.
The goal of the webinar is to present preliminary work on modeling complex disease trajectories, specifically for liver transplantation and pain (as well as some examples from previous work on sepsis) to capture patient deterioration while identifying interventions capable of prevent/delay such progressions. The webinar will also present several approaches for building different phenotyping strategies for these diseases as part of the process, validation methods, and the importance of team science for more accurate and clinically relevant.