Most healthcare AI does not fail because the model is weak. It fails because governance is weak, workflows are wrong, and no one owns accountability after go-live.

About Dr. Ryan Sadeghian

My path into clinical informatics was not a pivot. It was a progression.

I trained first as a pediatrician because I wanted to understand medicine from the bedside, the clinical reality, the documentation burden, the moments where the system either supports or fails the physician trying to care for a patient. That foundation never left. It is why my approach to AI and informatics is clinical first, not technology first.

I hold dual fellowship training in Clinical and Biomedical Informatics from the University of Washington and the University of Pittsburgh, with post-doctoral NIH-funded training in machine learning at the National Library of Medicine. That combination means I can evaluate an AI model's design, deploy it inside a real clinical environment, measure whether it changes outcomes, and defend it when it is challenged clinically or legally.

Healthcare AI is only as good as the clinical and operational infrastructure around it. I have spent my career building that infrastructure, and fixing it when it fails. Over the past decade I have founded a 55+ FTE Clinical Informatics Department, directed a $235M Epic transformation, deployed the first ambient AI documentation program in Northwest Ohio, and reduced enterprise payer denials by over 50%. I owned them, staffed them, and was accountable for their outcomes.

I currently serve as Enterprise Chief Medical Information Officer and Medical Director of AI at University of Toledo Health, Co-Chair of the HIMSS Physician Committee, and Founding Co-Chair of the inaugural HIMSS AI Committee. I am also a JD candidate focused on healthcare law and AI liability, because the legal and regulatory questions around clinical AI are no longer hypothetical. I am one of only eight professionals worldwide holding all three CHIME informatics certifications.