Waverly Rose Brim

Biography

Waverly Rose Brim is an AI research engineer building intelligent systems at the intersection of computer vision, biological foundation models, and surgical data science. Her work focuses on moving machine learning beyond the lab and into clinical environments where reliability, interpretability, and human decision-making matter. She develops production-grade AI systems that transform patient records, imaging studies, pathology, and clinical guidelines into auditable, decision-ready intelligence for care teams. Her current projects include SCRUB, a surgical computer vision platform designed to track gloved hands in the operating room where conventional pose-estimation models break down; an optimized brain-to-image reconstruction framework capable of running on consumer hardware; and methods for evaluating the safety, fairness, and clinical risk of medical prediction algorithms. Her earlier work spans multimodal computational pathology and genomics for meningioma characterization, AI-driven surgical skill assessment, operating room analytics, and DARPA-supported neural engineering efforts developing genetically encoded voltage indicators. She has authored more than 20 peer-reviewed publications in medical artificial intelligence and neuro-oncology and previously moderated the “Dr. AI” session at the World Economic Forum's Annual Meeting of the New Champions in Tianjin, China. Brim is currently pursuing an M.S. in Artificial Intelligence at Johns Hopkins University and is driven by a simple question: How do we build AI systems that clinicians can trust, patients can benefit from, and society can responsibly scale?

Sessions

1. Building Fairness Into the Data Supply Chain

   Wed, Nov 4, 10:20 AM – 11:20 AM CT

   Track Data & Clinical AI

   Bias in healthcare AI begins before training, embedded in decisions about missing data, labels, and success metrics. Waverly Rose Brim demonstrates how a clinical prediction model with strong global performance can still fail Black patients and the underinsured, then walks through a fairness-aware correction workflow. Attendees leave with a practical framework for equity-first dataset design and go/no-go deployment standards.