Lorenzo joined the Bayes4Health project at Lancaster from April 8^th 2021. He will submit his PhD thesis at the University of Bristol at the School of Mathematics in April 2021. The unifying theme of his research has been to overcome the prohibitive computational cost of high-dimensional hidden Markov models, with the outcome of three papers [1], [2], [3]. Specifically, during the COVID-19 pandemic, he has nurtured a deep interest on epidemiology and how statistics can guide the governments' control measures to avoid a widespread of the disease. This has resulted in a recent publication “Inference in Stochastic Epidemic models via Multinomial Approximations” [3], accepted by AISTAT2021. In this work, he proposed a low-cost approximate filtering-smoothing algorithm for compartmental models, frequently used in epidemiology, with application in forecasting the transmission rate of COVID-19 in Wuhan between December 2019 and January 2020. In Lancaster, Lorenzo will work alongside Professor Paul Fearnhead and Professor Christopher Jewell on developing novel scalable methods fordisease control, which are both mathematically well-motivated and available to the research community through open source softwares.

[1] L. Rimella and N. Whiteley. “Exploiting locality in high-dimensional factorial hidden Markov models”. ArXiv (arXiv:1902.01639v1 [stat.ML]). 2019.

[2] L. Rimella and N. Whiteley. “Dynamic Bayesian Neural Networks”. ArXiv (arXiv:2004.06963v2 [stat.ML]). 2020.

[3] N. Whiteley and L. Rimella. “Inference in Stochastic Epidemic models via Multinomial Approximations”. Arxiv (arXiv:2006.13700v1 [stat.ME]). 2020.