Reporting Apps for data projects

My plenary talk at the REDCap (https://www.project-redcap.org/) conference in NYC, August 2017. All apps are R-Shiny and code is available at https://github.com/riinuots/

Transcript

1. Reporting Apps Riinu Ots, PhD Data Manager:
2. Outline: ‣background/our setting ‣Examples of the reporting apps ‣How to install
3. Background: GlobalSurg studies ‣Use REDCap to collect data that is: ‣ Patient level ‣ Comprehensive - 3 instruments, 34 variables ‣ Crowdsourced ‣ Globally ‣For GlobalSurg 2 (1-Jan - 31-Jul 2016), this meant: ‣ 3,277 Registration Survey responses ‣ 2,127 REDCap users from 70 countries ‣ 554 hospitals (DAGs) ‣ 16,717 records ‣ REDCap statistics (24-Jul-2017): ‣ Edinburgh: 8,413 users in 126 Production Projects ‣ Vanderbilt: 32,217 users in 28,288 Production projects
4. Reporting and data quality: Built-in tools
5. Reporting Apps ‣Monitor data collection and Aid cleaning ‣Report Aggregated data back to collaborators ‣Help deliver an intervention with REDCap* *TWIST (tracking Wound Infection Using Smartphone Technology): A Randomised Control Trial where the intervention is delivered by REDCap+Shiny
6. Reporting Apps: Monitor data collection and Aid cleaning Code: github.com/riinuots/redcaptime Live at: https://riinu.shinyapps.io/redcaptime/
7. Reporting Apps: Report Aggregated data back to collaborators Code: https://github.com/riinuots/gs2_ssi_app https://github.com/riinuots/shinyviz Live at: ssi.globalsurg.org Empirical Bayes credible intervals: (Reproducible code to be released shortly.)
8. In action: ssi.globalsurg.org
9. no p-values??? No p-values.
10. Patient consents to take part in the TWIST Randomised Control Trial Control arm 30-day follow-up by a blinded clinician Intervention arm Patient’s phone: Clinician’s response Patient’s phone: Reporting Apps: Delivering an intervention
11. Developed with: Statistical Programming Language Integrated Development Environment (IDE) for R: ‣ web server - OR ‣ desktop Web application framework for R: ‣ web server ‣ desktop + + This relationship is somewhat similar to:
 MySQL -> REDCap -> JavaScript (built-in reporting tools)
12. Developed with: Statistical Programming Language Integrated Development Environment (IDE) for R: ‣ web server - OR ‣ desktop Web application framework for R: ‣ web server ‣ desktop + +
13. install.packages (“shiny”) + shinythemes, ggplot2, tidyr, dplyr, forcats, magrittr, scales (All tidyverse) shiny::runGitHub(“shinyviz”, “riinuots”) Instructions at: github.com/riinuots/shinyviz
14. shiny::runGitHub(“shinyviz”, “riinuots”) Instructions at: github.com/riinuots/shinyviz
15. ‣ Every time the app is opened ‣ Scheduled (Cron jobs) ‣ Based on a data entry trigger ‣“manual” update + = But how often should they talk? Really depends on the project, could be:
16. Thank you for listening! Get Code: github.com/riinuots R training: healthyr.surgicalinformatics.org Riinu Ots, PhD Data Manager R.Ots@ed.ac.uk