DrSara Fontanella

My research interests focus on the development of multivariate statistical models, particularly in a Bayesian framework, with applications spanning medicine, social sciences, and environmental studies. The main areas of investigation include: i) Machine learning and Multivariate Statistics for handling high-dimensional data, ii) Graphical models for manifold learning, iii) Analysis of longitudinal and functional data, iv) Latent variable models, factor models, and v) Item Response Theory models.

I completed my Ph.D. at the School of Advanced Studies G. d'Annunzio, Chieti (Italy), where my thesis explored "Graphical models for spectral nonlinear dimensionality reduction," aiming to uncover relationships within high-dimensional data.

In 2013, I began a postdoctoral position at the Open University, focusing on “Sparse Factor Analysis (FA)” in large datasets. My research involved innovative approaches utilizing L1 penalty terms and Bayesian statistics with Markov chain Monte Carlo methods, particularly beneficial for high-dimensional data analysis. Transitioning to Imperial College's Faculty of Medicine in 2016, I joined the Study Team for Early Life Asthma Research (STELAR) as a Research Associate in Statistical Machine Learning. Here, I implemented computational statistical methods to identify novel subtypes of childhood asthma. In 2019, I worked as a Research Fellow at the University of Torino (Italy), concentrating on statistical models for functional data analysis in environmental science, utilising Gaussian Processes for spatio-temporal data analysis.

Currently, I hold a position as a Non-Clinical Lecturer in Biomedical Modelling within NHLI at Imperial College London. My ongoing research focuses on implementing and developing computational statistical methods to understand the heterogeneity of allergic sensitisation and its role in respiratory diseases, enabling investigation into distinct pathophysiological mechanisms.