Design, modelling and analysis for longitudinal population studies involving high-dimensional molecular measurements

Longitudinal population studies often involve obtaining blood or tissue samples to measure molecular characteristics. Recently, high-throughput omics technologies have allowed us to collect large numbers of molecular measurements from a single sample. Analysing these large datasets is problematic because existing techniques do not allow us to analyse all measurements jointly as outcomes.

We will develop a statistical technique to jointly analyse large numbers of molecular measurements, using recent advances in statistical inference, as well as applying prior knowledge to reduce the number of relationships between measurements that need to be explored. We will also develop a method for designing longitudinal studies to gain optimal information about these high-dimensional outcomes.

Our approach will allow us to gain additional information about the relationships between longitudinal measurements and it will also help improve the design of future studies, which will lead to time and cost savings.