Loading Events

e-HAIL Event

e-HAIL Research in Progress: A Perioperative Point of Care Clinical Support System Leveraging AI

Michael Burns, Ph.D.Assistant Professor, AnesthesiologyU-M Medical SchoolMaggie Makar, Ph.D.Presidential Fellow in AI, Electrical Engineering and Computer Science U-M College of Engineering

While there have been remarkable advances in the safety and accessibility of global perioperative (pre-, intra-, and post-operative) healthcare, there remain large variations in care. Annually there are an estimated 313 million procedures performed worldwide – even small improvements in the provision of safe and affordable surgical and anesthesia care can have a major impact on public health.  Leveraging a multicenter perioperative electronic health record (EHR) registry, we have been able to use off-the-shelf artificial intelligence/machine learning (AI/ML) tools to identify and create hundreds of meaningful phenotypes to develop a provider-facing application to summarize variation in perioperative care. The coaggregation of diverse perioperative data from many hospitals allows us to identify naturally occurring variations that enable us to use quasi-experimental designs to better understand which care patterns yield the best outcomes in which patients. We have an opportunity to apply novel AI/ML and causal inference techniques to these unique perioperative data to learn individual treatment effects. In using point-of-care tools, we can provide an evidence-based application to inform best anesthesiology practices and improve quality of care, cost, and safety to help bridge quality and accessibility gaps across global healthcare in providing evidence-based clinical processes.


J. Henrike Florusbosch