Artificial intelligence has rapidly affected the healthcare landscape with applications ranging from prognostic models to language-based systems and emerging generative and agentic AI applications. But performant systems might not be clinically impactful. Drawing on pragmatic clinical trials of AI-driven interventions in urban and rural clinical settings, this talk will explore how silent validation, workflow integration, governance navigation, and continuous monitoring can bridge the gap between AI/ML development and improved patient care. The talk will also discuss how advances in language models and phenotyping are expanding what we can measure in healthcare, while reinforcing the need for rigorous evidence generation in real-world clinical settings.

Bio: Colin Walsh, MD, MA, FACMI, FAMIA, FIAHSI, is an Associate Professor of Biomedical Informatics, Medicine, and Psychiatry at Vanderbilt University Medical Center. His research focuses on developing, evaluating, and implementing artificial intelligence to improve healthcare delivery, with an emphasis on pragmatic clinical trials, clinical decision support, natural language processing, and mental health. Dr. Walsh leads multisite, multidisciplinary efforts spanning clinical prediction, multimodal phenotyping, and public health informatics, with the goal of generating rigorous evidence that AI can improve clinical decision-making, healthcare quality, and patient outcomes in real-world practice.