Pouria Rouzrokh, MD, MPH, MHPE

My research centers on the intersection of artificial intelligence and medicine, with a particular focus on medical imaging. I’ve worked extensively on developing AI tools for automating quantitative measurements in radiology, with much of my work during my research fellowship focused on hip arthroplasty. I’ve also explored using AI to curate large-scale datasets and build systematic imaging registries, enabling more structured and scalable research. In parallel, I’ve developed generative AI models for applications such as super-resolution imaging, 3D reconstruction from 2D scans, and forecasting a patient’s imaging trajectory based on prior studies. More recently, I’ve become increasingly involved in the use of large language models and multi-agent frameworks to streamline clinical and research workflows—including scientific writing, data synthesis, and literature reviews. One area I’m particularly passionate about is the explainability and fairness of AI models. My work investigating algorithmic bias—especially in a trilogy of studies published in Radiology: Artificial Intelligence—highlighted how data, development, and performance metrics each introduce unique vulnerabilities to bias in medical AI. Alongside this technical research, I’ve also authored review articles aimed at translating complex AI concepts into accessible language for clinical audiences. I have contributed to multiple clinical fields, including musculoskeletal imaging, neuroimaging, and general clinical data science.