Priyanka Kopuru
Salah Mohamed
Alex Thomas
Current Electronic Health Records contain rich longitudinal which is important to make medical decisions but current methods aren't able to synthesize and retrieve on temporal data. Our approach uses RAG model with temporal reasoning to improve response relavence over time and provide accurate results in cardiac care
What did you try to do? What problem did you try to solve? Articulate your objectives using absolutely no jargon.
Electronic Health Records (EHRs) contain rich longitudinal data critical for clinical decision-making. However, most NLP systems struggle to reason over time—especially when events span days, weeks, or months. Retrieval-Augmented Generation (RAG) models offer promise by combining retrieval with generation, but current systems lack temporal awareness. Our goal is to enhance RAG models with temporal reasoning to improve response relevance and accuracy in cardiac care scenarios. This could support clinicians in understanding patient trajectories, treatment timelines, and symptom progression.
How is it done today, and what are the limits of current practice?
Currently, the main method is to have doctors review patient records over time, and this is overall more time consuming.
Who cares? If you are successful, what difference will it make?
This would help in making it faster to synthesize patients' records for whoever might find it useful