MaryLena Bleile, Ph.D

Statistician

Welcome to my website! I am a Canadian statistician currently living in New York City. I am interested in the design of Bayesian and mixed models, as well as [Causal] Reinforcement Learning, which I mostly apply in the context of biomarker development using clinical -omics data. My current role is in pharmaceutical support using AI for biomarker development at Sanofi. Past contracts with Memorial Sloan-Kettering Cancer Center (postdoc) and UT Southwestern Medical Center's AI and Automation lab (Ph.D) have been particularly pleasant.

I believe that one of the major problems in the world is a phenomenon called dichotomistic bias: The human tendency to simplify phenomena by categorizing and imposing dichotomies. Dichotomies aren't necessarily bad in themselves - indeed, every statistical model encodes a dichotomy of the variability in the data into "signal" and "noise" - however, our tendency to simplify by imposing them when they aren't useful often gets in the way of justice, honesty, and good scientific research. Moreover, the effects of this dichotomistic cognitive bias not only  permeate statistical practice, but also extend to the broader elements of life, driving the cartesian dualistic framework of society in which we currently live.  

I am committed to dismantling the grip of dichotomistic bias by counteracting false dichotomies with balance and multiplicity; allowing for multiple perspectives, models, or statistical frameworks instead of revering just one as the sole proprietor of truth. In the words of Francois Chollet, "I want to write code that feels like art, and make art that thinks like code". 

 

When I'm not doing statistics, you can usually find me on the mat doing Brazilian Jiu Jitsu or Judo (watch one of my competition matches here), or lifting weights at the gym. 

Photography by Anthony Batista