Statistical causal modelling approaches to understanding the developmental profiles of asthma and allergy

The past decade has seen rapid advances in big data mining, with clustering of longitudinal data used to identify a variety of developmental profiles of asthma as well as to disaggregate allergic sensitisation. This project will aim to further the progress in this area by moving beyond current modelling approaches and focusing on data integration across different variables, from patient reported outcomes to blood biomarkers and lung function measurements.

Main data sources for this project will be five population based birth cohorts that form part of the Study Team for Early Life Asthma Research (STELAR) network, which is described in detail elsewhere. Collectively, these cohorts provide data for > 14,000 children and have similar data structures that will facilitate analysis. At each time point, validated questionnaires were used for gathering information on the occurrence and frequency of wheezing, atopic eczema and rhinitis. IgE responses to different allergen components and various lung function measures were also recorded at some of the time points.

A major goal of the project will be the discovery of clinically relevant phenotypes that are homogeneous and share underlying pathophysiological mechanisms. While previous studies have used data-driven approaches to derive phenotypic clusters, there is a pressing need to refine these by integrating a broader range of the data available to us, and make sure they relate to outcomes and demonstrate clinical utility. The project will aim to replicate findings across birth cohorts, and potentially verify them via disparate lines of evidence such as data from neonatal mouse models in later stages.