An Open-Source "R-Package" and "R-Shiny" Application Designed to Integrate Single-Arm Observational Data and Bridge Gaps in Disconnected Evidence Networks

OBJECTIVES: Network meta-analysis (NMA) enables the estimation of comparative effectiveness for treatments that have not been directly compared in head-to-head trials. However, the relative treatment effects for all interventions can only be determined if the available evidence forms a connected network. A method detailed in Schmitz et al. helps bridge disconnected networks by leveraging covariate information from single-arm observational studies. It evaluates the similarity between studies by considering patient characteristics and calculating a distance matrix. The objective of this study was to develop an open-source R package to standardize the application of the methodology, publish it on the Comprehensive R Archive Network (CRAN), and create an intuitive "R-shiny" application for non-programmers.

METHODS: The dynamic functions of the R package were developed using the "roxygen" framework. Three essential functions include "check_data," which validates the imported data to ensure it conforms to the specified format to prevent errors during analysis; "specify_weight," which assigns weights to different covariates; and "calc_dist," which calculates the distance matrix between studies. The user-friendly web application, developed with the R Shiny framework, is deployed using Docker containers on Amazon Web Services (AWS) to ensure easy access and scalability. User security is maintained with SSL certificates and Auth0 authentication. For data security, uploaded data is temporary and not stored after the session ends.

RESULTS: The R package "closeloop" was successfully deployed on CRAN and GitHub, allowing users to install it on their local computers. Additionally, an R Shiny tool was developed for interactive data inputs and outputs from the package and successfully deployed on AWS with Auth0 authentication.

CONCLUSIONS: The developed open-source R package and R Shiny tool effectively standardize and simplify the application of Schmitz et al.'s method by leveraging covariate data to connect networks, enhancing user accessibility to non-programmers. It empowers users to conduct sophisticated analyses with ease and confidence.