Kajetan Stanski

I am a PhD student at the Roslin Institute, University of Edinburgh funded by EPIC and supervised by Mark Bronsvoort, Sam Lycett and Thibaud Porphyre. In 2017, I did my integrated MSc in computing science and physics at the University of Glasgow with my thesis on machine learning techniques for classification of Higgs boson events. I worked as a research assistant at the Aalto University in Finland where I was responsible for developing data-driven models predicting physical properties of chemical compounds.

Within EPIC, I contribute to improving veterinary surveillance strategy by applying machine learning methods to animal surveillance data. In Scotland and the rest of GB, surveillance data (e.g. screening test results) are recorded in databases and made available to EPIC researchers. What makes machine learning incredibly useful, is an automatic discovery of patterns and correlations within data. This means that reliable predictions can be made even if a disease spread dynamic is not fully understood or if infection risks are hard to quantify.

In 2018 and 2019, I developed a machine learning model capable of accurately predicting herd-level bovine tuberculosis breakdowns in GB. This model captured information from multiple data sources including breakdown history, between-farm cattle movements and land cover data to help target high-risk farms for follow up testing. Currently, my focus is on improving the pre-processing of cattle movement network to enable the model to take full advantage of this rich dataset. I believe predictive machine learning models can aid disease control activities and support better-informed decision making.


Using machine learning improves predictions of herd-level bovine tuberculosis breakdowns in Great Britain K. Stański, S. Lycett, T. Porphyre & B. M. de C. Bronsvoort

Website design by Innovation Digital Limited