Outcomes-based data science to optimise individual therapy choice in type 2 diabetes

Optimisation of glucose-lowering treatment in type 2 diabetes is important to reduce the risk of patients developing microvascular and macrovascular complications. Despite the known heterogeneity of type 2 diabetes, and the multiple drug classes available with different mechanisms of action, treatment decisions do not take into account the clinical characteristics of individual patients. The aim of this study is to improve clinical outcomes in type 2 diabetes by developing a universal approach to optimise therapy choice for individual patients based on their characteristics. To achieve this, I will apply a novel outcomes-based data science approach to conduct studies using routine clinical data (discovery) and individual-level data from clinical trials (validation). This will allow the discovery of patient characteristics that alter clinical outcomes (cardiovascular and renal disease, glycaemic response, weight, and tolerability) on all major type 2 diabetes therapies after metformin, and the development and validation of treatment selection algorithms to provide individualised estimates of risk/benefit with each medication. Reproducibility will be tested in people of different ethnicities to assess generalisability. Studies will prioritise routine clinical features meaning the approach developed is low-cost, with the potential to lead to the development of a practical universal approach to treatment individualisation applicable worldwide.