CAR.L.L.: An LLM Agent for Physiologic Inversion

Daniele Schiavazzi and members of the CARLL team — Daniele Schiavazzi, center, is the principal investigator of the CAR.L.L. team. He is pictured here with, from left to right: Margaret Nichols, data scientist at the Lucy Institute for Data & Society; Elizabeth Hanley, undergraduate research assistant; Shifu Hou, postdoctoral research associate at the Lucy Institute and Grayson Tong, doctoral student researcher. Not pictured: co-PIs Nitesh Chawla and Fang Liu.

Head PI: Daniele Schiavazzi, associate professor in the Department of Applied and Computational Mathematics and Statistics

Co-PIs: Nitesh Chawla, the Frank M. Freimann Professor of Computer Science and Engineering at the University of Notre Dame and the founding director of the Lucy Family Institute for Data and Society, and Fang Liu, professor in the Department of Applied and Computational Mathematics and Statistics

CAR.L.L. is an advanced AI agent designed for monitoring and analyzing cardiovascular health. It can estimate quantities such as blood pressures, flows, volumes, and heart performance in real-time by integrating clinical assessments, wearable sensor data, and socio-demographic information.

Unlike systems that rely solely on large datasets, CAR.L.L. leverages computational models based on physics. This approach ensures more accurate and fair predictions, making it equally effective for both well-represented and underrepresented social groups. Additionally, CAR.L.L. enhances the capabilities of Large Language Models (LLMs, like ChatGPT, Llama, Claude, etc.) by improving their understanding of cardiovascular physiology. This enables more meaningful and informed interactions in healthcare conversations. Protecting user privacy is a core principle, with CAR.L.L. specifically designed to safeguard sensitive health data.