sustainable outbreak analysis ecosystem with recon

The R Epidemics Consortium (RECON) builds and maintains a collection of interconnected software tools focused on actionable analyses to be used in disease outbreak response. These include several kinds of tools written in the widely used R language. "Building block" software defines data structures and processing steps that can be used by a wide set of other tools and epidemiological pipelines, e.g., epicontacts, for contact tracing data, or distcrete for working with discrete distributions in disease simulations. There is also "turnkey" software for low-effort, rapid implementation of common analyses required in outbreak response, e.g.  EpiEstim, for calculation of reproductive numbers, and epiflows for travel-based risk assessment. RECON also build tools for analytical support to low-resource environments such as deployer, a framework for bootstrapping tools without internet access, and reportfactory, for rapidly generating standard epidemiological reports. Developed since 2016, and deployed and tested through multiple Ebola outbreaks and the COVID-19 pandemic, RECON tools are essential components of rapid response analytics. RECON software has been built by a loose coalition of volunteer developers with a mix of software design approaches. As the R epidemiological toolset has expanded and developed, other ecosystems have been built using RECON packages as essential dependencies, while other specialty epidemiological tools have developed parallel but incompatible data structures. This project aims to bring RECON packages in line with each other and peer tools in terms of package design standards so as to ensure dependability and maintainability. We will bring the core RECON suite of packages up to common set of standards in testing, documentation, and maintainability in design following standards for R and epidemiological software, as defined by the rOpenSci development guide and Epiverse-TRACE blueprints. Core RECON packages with significant reverse dependencies and user bases will be the priority, including EpiEstim, outbreaks, epicontacts, distcrete, incidence2 and i2extras.  Priority maintenance within these include updating core algorithms with updated methods (distcrete), improving API for interoperability with emerging popular packages (epicontacts, incidence2, EpiEstim), and refactoring code to facilitate further expansion (EpiEstim).  All will receive updates to improve test coverage and developer documentation on package internals and roadmap to improve maintainability, facilitate regular release cycles to dissemination platforms such a the Comprehensive R Archive Network (CRAN) and R-Universe,  and expand the contributor base. We expect to responsibly deprecate several packages that have low usage and have been superceded by new tools. As part of this effort, we will coordinate with Epiverse, a peer project using RECON tools, to align on standards and establishing broad, long-term maintenance priorities. To improve both internal and external transparency into package maintenance status, we will implement monitoring for all RECON packages using the Community Health Analytics in Open Source Software (CHAOSS) framework.  A CHAOSS dashboard will provide insight into RECON package development cadence, developer responsiveness, and package cross-dependencies with the broader software ecosystem.  The RECON developer board will use the CHAOSS dashboard to target packages for additional developer support when falling behind on maintenance.