AI-Enabled Sepsis Decision Support: What Works, What Doesn’t, and What We Should Focus on Now
Description:
This session traces the evolution of computer‑assisted sepsis detection, from early predictive models to today’s advanced clinical decision support systems. Drawing on two decades of real‑world experience, the presenter will examine why high‑performing algorithms have not always led to better patient outcomes and unpacks the practical challenges that shape success, including workflow integration, clinician trust, alert fatigue, bias, and governance. Attendees will gain a stronger understanding of how these lessons are redefining what responsible, effective AI in clinical care must look like.
Learning Objectives:
At the end of this session, the learner should be able to:
- Describe the limitations of AI-based sepsis prediction models when applied in real-world clinical settings, including factors that affect clinical adoption and patient outcomes.
- Apply principles of decision-aware clinical decision support to evaluate AI-assisted sepsis tools within existing clinical workflows.
- Assess organizational, ethical, and governance considerations relevant to the implementation and oversight of AI-enabled sepsis decision support in healthcare systems.
Target Audience:
Nurses, advanced practice providers, physicians, emergency responders, pharmacists, medical technologists, respiratory therapists, physical/occupational therapists, infection prevention specialists, data/quality specialists, and more.
Vitaly Herasevich, MD, PhD, MSc
Professor of Anesthesiology; Professor of Medicine in the Department of Anesthesiology and Perioperative Medicine
Division of Critical Care, Mayo Clinic - Rochester
Vitaly Herasevich, MD, PhD, MSc, is a Professor of Anesthesiology and a Professor of Medicine in the Department of Anesthesiology and Perioperative Medicine, Division of Critical Care at the Mayo Clinic, in Rochester, Minnesota. He has been involved in medical informatics for over 25 years, with a specific concentration on applied clinical informatics in critical care and the science of health care delivery.
Dr. Herasevich joined Mayo Clinic in 2006 and co-directed the Mayo Clinic Clinical Informatics in Intensive Care Laboratory. Dr. Herasevich's applied clinical informatics in critical care work included clinical data representation, ambient decision support, and alerting systems for early detection of critical syndromes and hospital-wide surveillance. He architected data warehouses in support of clinical decision-making quality and outcomes research. With longstanding interest in information security and more recent interest in computer vision, Dr. Herasevich has coauthored over 150 scientific articles and >span class="NormalTextRun SCXW247208915 BCX0">book “Health Information Evaluation Handbook” (now in the second edition).
Dr. Herasevich has a long mentorship record. In addition to being appointed with full faculty privileges in clinical research and artificial intelligence at Mayo Graduate School, he developed the Informatics Curriculum for Mayo Medical School, CCaTS, and the Master’s in Medical AI programs. Dr. Herasevich is principal investigator, co-investigator, and informatics expert for past and ongoing federal and industry-funded projects totaling more than $115 million in research support. Dr. Herasevich is Fellow of American College of Critical Care Medicine (FCCM), Fellow of HIMSS (FHIMSS), and President of Minnesota HIMSS Chapter.
No relevant financial relationships to disclose.