ESPE Abstracts

Background: Rare disease registries provide rich longitudinal data on patients with rare conditions that can be used for research. Those charged with access to data must safeguard the process, but also have a good oversight of the available data to be able to work with researchers to realistically assess whether there will be adequate data available to answer specific research questions. Data access committees would benefit from a pragmatic method of rapidly assessing the amount of data and relationship between variables available to support study design and approval.

Aim: We set out to develop a web application that provides an interactive display of available registry data.

Method: A generative adversarial network (GAN) was used to create mock longitudinal data from an example dataset extracted from the International Congenital Adrenal Hyperplasia Registry (I-CAH) that allowed demonstration of the application. R within RStudio was used to produce a Shiny web application that could provide rapid visual representation of the data. We analysed run performance using Chrome DevTools and conducted preliminary user testing to assess speed of use and improve the user interface.

Results: The application developed is available at disease-registries.shinyapps.io/DataVisualiser/ and provides the user with 3 different visualisations after uploading the mock dataset. The first allows visualisation of continuous variables, the second summarises categorical variables, and third allows identification of outlying points that warrant interrogation to assess data entry error. The app demonstrated robust performance, with time to interactive of 0.56 seconds on high-speed broadband. User testing with five users showed all able to load data and generate an informative output. Time from loading to generating the output ranged from 18 to 50 seconds. Qualitative enquiry showed good feedback from all users, and provided important suggested improvements now incorporated into the user interface.

Conclusion: The amount of data available within disease registries benefits from up to date summary and visualisation so that its quality and quantity can be assessed by data access committees charged with the approval of research data requests. Applications such as the one developed in this project can support faster and more informed assessment of rare disease registry data access requests.