Yirui (Iris) Hu's work focuses on developing and validating clinical risk prediction models using large‑scale electronic health record data across multiple healthcare systems. As a co‑investigator, she has led multi‑site model development and validation efforts through the PsycheMERGE Network. Her research emphasizes advancing adaptive modeling approaches, including transfer learning, to improve model performance and generalizability across heterogeneous sites, as well as developing federated learning frameworks that enable collaboration without sharing individual‑level data. Through this work, she seeks to translate methodological innovations into clinical practice by enabling early identification of high‑risk patients, supporting targeted interventions, and informing clinical decision‑making to improve patient outcomes.