Biography
Thomas Bury is an applied mathematician, data scientist, and TEDx speaker. He currently works at the Department of Physiology, McGill University. His research uses tools from mathematics and artificial intelligence better understand the dynamics of complex systems arising in ecology, epidemiology, climate science and medicine. He is currently working on methods to detect and classify cardiac arrhythmia in people. Besides research, he writes on Medium for Towards Data Science and teaches a range of courses.
- Machine learning
- Dynamical systems
- Mathematical modelling
- Responsible AI
- Open source
PhD in Applied Mathematics, 2020
University of Waterloo, Canada
MMath in Mathematics and Theoretical Physics, 2015
University of Cambridge, UK
BA in Mathematics, 2014
University of Cambridge, UK
Experience
- Partnered with cardiologists (UBC, Cornell) and industry (Icentia inc.) to research mechanisms of cardiac arrhythmia from wearable device data.
- Pre-processed ~60GB of electrocardiogram data from 12k patients using cloud resources (Compute Canada).
- Created and deployed data dashboards (Plotly, Dash) to visualise records and derived metrics.
- Implemented unsupervised machine learning (k-means) and dimension reduction techniques (PCA+t-SNE) to investigate arrhythmia subtypes.
- Awarded a FRQNT Postdoctoral Research Scholarship ($45k per annum).
- Supervised 1 graduate and 2 undergraduate students. Project trained a reinforcement learning agent in Tensorflow to generate an arrhythmia with minimal stimulus.
- Course: “Foundations of Quantitative Life Sciences”.
- Co-instructed 15 graduate students. Co-managed 1 teaching assistant.
- Consultant for mathematical modeling and analysis of birdsong.
- Created visualisations for the book “Parasitic Oscillations” by Dr. Madhur Anand, published by Penguin Random House.
- Thesis: “Detecting and distinguishing transitions in ecological systems: model and data-driven approaches”.
- Developed methodology and an open-source Python package for detecting tipping points (bifurcations) in time series data.
- Tested methodology with climate, ecological, geological and engineering datasets, which outperformed conventional methods.
- Published as first-author in leading journals including PNAS, Journal of the Royal Society Interface, and PLOS Comp. Biology.
- Presented work at international conferences including TEDx, Soc. for Mathematical Biology, and Canadian Soc. of Applied and Industrial Mathematics.
- “Calculus I for the Sciences”. 115 undergraduate students, 1 teaching assistant.
- Received strong student evaluations (>4.5/5 average for each teaching aspect).
- Processed and analysed disease incidence data from hospitals across East England.
- Presented findings to public health professionals at Addenbrooke’s Hospital, Cambridge.
Selected publications
Public talks
TEDx conference
I was invited to a TEDx conference at the University of Toronto, where I gave an accessible overview of tipping points and how mathematics can help us understand and detect them.
Centre de recherches mathématiques
A more technical presentation of my work on using deep learning to predict tipping points.
Three-Minute Thesis, University of Waterloo
A short summary of my PhD research at the finals of the Three-Minute Thesis competition.
Contact
- thomas[dot]bury[at]mcgill[dot]ca
- McGill University, 845 Sherbrooke St W, Montréal, Québec H3A 0G4