Healthcare data is abundant, representing approximately 30% of the entire global data volume,1 and is becoming increas- ingly available to researchers to allow for such interrogation as trend analysis, pattern recognition and predictive mod- elling. This is helped primarily by the increased adoption of electronic health record (EHR) systems in hospitals, with most UK NHS Trusts currently using one and all expected to have one by 2025.2 In parallel, there has been a revolution in the capabilities of machine learning (ML) methods, allowing for the efficient analysis of high-dimensional clinical and imaging data.