Abstract
ObjectiveA framework and toolbox for creating point-and-click dashboardapplications (at no cost) for monitoring several facets of syndromicsurveillance data was created. These tools (and associateddocumentation) are being made available freely online for othersurveillance practitioners to adopt.IntroductionPublic health surveillance largely relies on the use of surveillancesystems to facilitate the identification and investigation ofepidemiologic concerns reflected in data. In order to support publichealth response, these systems must present relevant information, andbe user-friendly, dynamic, and easily-implementable. The abundanceof R tools freely-available online for data analysis and visualizationpresents not only opportunities, but also challenges for adoptionin that these tools must be integrated so as to allow a structuredworkflow. Many public health surveillance practitioners do not havethe time available to 1) scavenge for tools, 2) align their functionsso as to create a relevant set of visuals, and 3) integrate these visualsinto a dashboard that allows a streamlined surveillance workflow.An openly-available, structured framework that allows simpleintegration of analytic capabilities packaged into readily-implementable modules would simplify the creation of relevantdashboard visuals by surveillance practitioners.MethodsR is a statistical computing application, known for its versatilityand ability to create powerful visualizations. Shiny is an R packagethat allows the creation of interactive, easy-to-use point-and-clickapplications. We looked to R and its Shiny package extension asa candidate solution. However, creating a Shiny application fromscratch requires knowing enough of the R programming languageso as to be able to appropriately design and link several chunks ofcode that interact with one another to generate the desired output.To address this barrier, we sought to create a structured processby which one can easily browse a library of defined code snippets(each of which enables an analytic tool relevant to syndromic dataanalysis and visualization) and then integrate snippets of interest intoa dashboard application in a way that requisite experience with R isminimized.ResultsWe first collected several analytic tools that support syndromicdata analysis and have been developed for R; examples includeheatmaps, change-point detection, outlier detection, tables, maps, etc.We then packaged them into snippets of code (one for each analytictool) in a way that facilitates integration of the analytic tool into adashboard application. A fake syndromic dataset was created as wellfor inclusion in a demo dashboard application that is available forsharing.ConclusionsThe online community of R users makes new tools for data analysisand visualization available every day. The abundance of options canbe overwhelming and the process of integrating pieces of code canbe time-consuming. This places a constraint on adoption of thesetools by epidemiologists working at all levels of government. Thepresent project alleviates this problem considerably by reducing thetool searching process through the introduction of a library of relevanttools for syndromic data analysis and visualization that can be easilyintegrated into a dashboard application that allows for streamlinedsyndromic surveillance activities.Our next step is to partner with interested jurisdictions to help themadopt this framework and associated tools. Given sufficient interest,we would set up a process for others to add their own modules to thislibrary, perhaps through the online platform for collaborative codedevelopment and sharing, GitHub.