Using unsupervised computational methods on electronic health records, to identify changes in underlying infrastructure or data quality over time

Electronic health records (EHR) are increasingly being used to conduct epidemiological research but there is little awareness or guidance regarding appropriate methods to check the validity of using such data when it spans long time periods.  Many things which contribute to how and what data is recorded, such as the reorganisation of services, or updates in software, will likely have occurred in hospitals over time, and this study will use modern computer-based methods to attempt to identify these types of changes quickly and easily, and with minimal human intervention.  It will assess what the most effective methods are, by comparing them to each other as well as to known changes and updates at the OUH, and it will create a tool that is readily available to other researchers so that they can use it to improve the quality of their own studies.