In the fight against cancer, AI can be an extremely powerful tool for both clinicians and patients, but the high stakes require nothing short of excellence in operation and ease of use. Researchers and developers have become increasingly aware that transparency is key to both; when a model or agent \u201cshows its work,\u201d human users are able to both verify and trust its accuracy.<\/span><\/p>\n Professor Kayhan Batmanghelich (ECE) has considerable experience working on both the clinical accuracy and reliability of AI tools for healthcare, from <\/span>addressing biological and demographic biases<\/span><\/a> in medical AI to <\/span>creating powerful new models<\/span><\/a> capable of subtyping and predicting particularly challenging diseases. Most recently, he\u2019s contributing this expertise to a new multi-institutional collaborative effort to build advanced AI models for predicting breast cancer recurrence and prognosis.<\/span><\/p>\n With a combined $4M in support from the NIH\u2019s National Cancer Institute (NCI), Batmanghelich and colleagues including PI Joann Elmore (UCLA), Linda Shapiro (University of Washington), and researchers from several other organizations will utilize large-scale datasets drawn from clinical trials and real-world patient cohorts to develop highly accurate, interpretable vision-language models for improved breast cancer survival predictions. The study will put particular focus on a pathologist-in-the-loop approach, where machine learning systems work collaboratively with users such as pathologists and clinicians to ensure clinically useful, transparent and trustworthy insights for personalized patient care.<\/span><\/p>\n
![\"\"](\"\/eng\/files\/2026\/04\/ece.speaker.26.batmanghelich.jpg\")
Assistant Professor <\/span>