A Shiny application developed by the Industrial Hygiene team of the Department
of Environmental and Occupational Health at the
École de Santé Publique of the
Université de Montréal (EPSUM) in collaboration
with Ununoctium (Jean-Mathieu Potvin).
Ununoctium designed the user interface (for versions
greater than or equal to 4.0.0) and currently maintains the web application
(but not scripts stored in scripts/).
This tool interprets a dataset of exposure measurements (including non detects) with regards to an OEL (Occupational Exposure Limit). In addition to multiple illustrative graphs, it exposes five components.
- Goodness of fit with respect to the lognormal model (graphical evaluation).
- Descriptive statistics.
- Risk assessment based on an exceedance fraction.
- Risk assessment based on percentiles.
- Risk assessment based on the arithmetic mean.
Calculations are performed using a Bayesian model fit using a Monte Carlo Markov Chain (MCMC) engine. It assumes that the underlying exposure distribution is lognormal.
The underlying Bayesian models and data interpretation procedures are derived from best practices in industrial hygiene data interpretation techniques. They are thoroughly described in Expostats: A Bayesian Toolkit to Aid the Interpretation of Occupational Exposure Measurements (Annals of Work Exposures and Health, Volume 63, Issue 3, April 2019, Pages 267–279).
To serve Tool 1 locally, call
.run()This sources .scripts/run.R. See this script for details.
Tool 1 is deployed to shinyapps.io.
To deploy a new version, call .pub().
# Deploy a beta version (useful for testing purposes).
# It is publicly accessible at https://lavoue.shinyapps.io/tool1-beta/.
.pub()
# Deploy an official version.
# It is publicly accessible at https://lavoue.shinyapps.io/tool1/.
.pub("prod")This sources .scripts/publish.R. See this script for details.
Some environment variables are required to publish new releases with .pub().
They must be stored in a top-level .Renviron file as shown below. This file is
ignored by Git and rsconnect.
# .Renviron
RSCONNECT_ACCOUNT_NAME=<account>
RSCONNECT_ACCOUNT_TOKEN=<token>
RSCONNECT_ACCOUNT_SECRET=<secret>
We are actively looking for external collaborators who could help us with supporting more languages. If you are interested, please send an e-mail to jerome.lavoue@umontreal.ca.
Tool 1 relies on package transltr
to support multiple languages. Translations (and related metadata) are stored
in intl/. To update its content, call
.find()This sources .scripts/find-text.R and updates all underlying
translations files. See that script for details.
Tool 1 further supports translation of ordinal numbers. Each supported
language requires an ordinal_rules_<lang>() function. For example, the
ordinal_rules_english() function implements grammar rules for English
ordinal numbers. See R/helpers-translate.R for more information.
Package transltr currently lacks
a proper way to (intelligently and incrementally) update existing translations
files. This feature will (likely) be released soon. Until further notice,
updating intl/ should only be done by
Jean-Mathieu Potvin.
Tool 1 uses sprintf()-placeholders (conversion specifications beginning by
%) to dynamically insert HTML content into template strings. Tokens such as
%s, %i, and %% in the source text and translations must be left as is.
See R/helpers-html.R for more information.
Tool 1 uses as few custom CSS (Cascading Style Sheets) rules as possible. It
does so by maximizing usage of bslib features and predefined
Bootstrap 5
CSS classes. Any CSS rule not declared in file www/main.css is highly likely
to stem from Bootstrap 5.
This is a well-known and well-maintained framework that can be trivially
understood and leveraged.
It is worthwhile to note that it is not necessary to include
Bootstrap's assets
(in the <head> of Tool 1) because bslib already does that automatically.
We may work on these issues in a near future.
-
French translations are officially supported but missing [COMING SOON].
-
Inputs lack robust validation mechanisms and may lead to undefined behavior in some cases.
-
Accessibility mechanisms (ARIA) are not implemented.
-
Margins, aspect ratios, font sizes, font families, dimensions, and colors of all plots are inconsistent and must be standardized.
- This could be achieved with
shiny::getCurrentOutputInfo()andbslib::bs_current_theme().
- This could be achieved with
-
In the source text, many inputs have slightly different names. For example, OEL (Occupational Exposure Limit) is sometimes named EL (Exposure Limit).
-
There are currently no explicit Terms of Service and Privacy Policy.
You may submit bugs, request features, and provide feedback by creating an issue on GitHub.
If you do not have a GitHub account, send an email to jerome.lavoue@umontreal.ca and jeanmathieupotvin@ununoctium.dev. Please use the following standard subject line:
Expostats [Tool 1] [Bug|Feedback]: <short description>