Presenter bio:
Byron Smith went to school at the University of Toronto before heading to Tennessee to earn a concurrent Masters in Statistics and PhD in Chemical Physics. He then taught at Virginia Tech for a year through the Academy of Integrated Sciences covering coursework from introductory calculus to graduate level machine learning. Byron then joined Mayo in 2015 as a statistician mostly supporting solid organ transplant research. His interests include the responsible use of Machine Learning and AI in medical research as well as how to leverage backgrounds like physics or other fields to do better research. Byron lives in Rochester, Minnesota with his wife, two sons, and dog.
Abstract:
Artificial intelligence is increasingly used everywhere from marketing to research and its use is becoming increasingly polarizing. There seems to be pressure from academic and business executives to apply these methods to streamline and automate processes while improving predictability. These methods can be quite powerful when used properly, but what does it mean to do it the right way? In this talk, I give some background in time-to-event methods that I have used as well as machine learning alternatives. I also discuss what may go wrong if these methods are used irresponsibly with a couple of examples from my own work.