David M. Kent, MD, MS Kicks Off the
Opening Plenary on Tuesday, April 9!

David M. Kent, MD, MS is Founder and Director of the Tufts Predictive Analytics and Comparative Effectiveness (PACE) Center, at the Institute for Clinical Research and Health Policy Studies (ICRHPS), Tufts Medical Center; Director of the Clinical and Translational Science (CTS) MS/PhD Program, at the Graduate School of Biomedical Sciences, Tufts University; and Professor of Medicine, Neurology, and CTS at Tufts Medical Center/Tufts University School of Medicine.

Dr. Kent is a clinician-methodologist with a broad background in clinical epidemiology, and a focus on advanced methods of predictive analytics and comparative effectiveness. He has been continually funded as Principal Investigator by both the NIH, since 2003, and PCORI, since their first round of research grants in 2012. He has received over $30 million from these agencies, both for work addressing fundamental analytic issues, and for applied work in cardiovascular disease.

Dr. Kent has over 250 articles in peer reviewed journals, including over 20 in the top 5 general medical journals (NEJM, JAMA, Lancet, BMJ, Annals of Internal Medicine). He has served on numerous boards, including the Scientific Advisory Board at Optum Labs, and the NIH, including Biomedical Computing and Health Informatics (BCHI) and the National Institute for Neurological Disorders and Stroke.

Presentation title: An Overview of Bias and Fairness in Algorithmic Models

Dr Kent will provide an introductory overview to the complex topic of Algorithmic Bias and Fairness. He will discuss the distinction between Bias and Fairness, and review different "fairness criteria" and the impossibility of simultaneously satisfying these. He will introduce the concept of "Label Bias" and present some examples related to clinical prediction models in which label bias may arise. Finally, he will discuss the use of race in clinical prediction models, and discuss the harms that may arise when race is omitted in some situations where the variable carries substantial prognostic information, and the necessity of evaluating trade-offs in considering its use in clinical prediction.