rhoinc / aeexplorer Goto Github PK

Specifying a list of group levels to be displayed (["Placebo","Treatment A"]) throws an error as shown.

Hopefully this is just buggy error checking and the functionality actually works ok.

From Jack's list. It would be helpful to the the patient's age when looking at AEs

When showing Jack the new SE in Prod, we noticed that once we added sex and race to the groups, the header colors no longer matched the values below it.

Not included in 3.0. Supported pre-3.0 with group:"All", but probably better to do it as an option now that we have the totalCol setting.

Biggest remaining hurdle is the settings.groups object, but that can be generated automatically. The chartfoundry renderer already does something similar.

Filter by subgroup, (sex, median age split, and active vs. placebo when unblinded)

Current behaviour: If a user specifies filters:null the default filters are rendered.
Expected behaviour: If a user specifies filters:null no filters are rendered.

Side notes:

• null is probably a more resonable default for filters
• filter:[] doesn't render any filters (this is fine)

cc: @bzkrouse

While customizing filters, I added a filter for sex and it flipped the colors and didn't give the new filter a header.

You also asked me to put "the code under the tech specs button" in here as well.

noticed by @srosanba

Defaults in AE explorer are specified as [], while defaults in some other safety widgets are 'null'. Convert all defaults to 'null' to make consistent and to make more sense logically.

Throw a console.warn() instead.

Right now, the AE table requires D3 v3.3.6 or earlier due to a change in d3.merge() after that release.

And then implement a build process using node/npm/rollup/babel similar to the one described in webcharts-wrappers-boilerplate

Pull this out of .init()

Not sure the best way to handle this, but it would be good to have the number of relevant study participants shown somewhere on the chart. Maybe add a footnote for "X of X participants shown in this view"

The text that shows up with the max groups error seems to be hardcoded and is currently incorrect: minor fix

The tool requires placeholder rows for participants with no AEs. Those participants are identified by a custom keyword in the user-specified major and minor column. Currently the keywords are hard-coded in /src/chart/util/prepareData.js. We should move them to defaultSettings.js instead and let the user overwrite the default as desired.

Per @srosanba , the file is simply named 'true' without an extension. Perhaps name it '[dataset name]-[group variable name] Summary' to give it a useful context.

Current options are not in intuitive order, consider ordering by:

• User preference
• or default to: Not related, Unlikely related, Possibly related, Related

Merge the current chartfoundry build process in to this repo

AETable.draw() probably makes more sense ...

Currently, the tool only supports filters on adverse event characteristics (e.g. severity). We need to bite the bullet and implement the ability for users to filter on characteristics that are related to participants (e.g. gender) as well.

Update chart arguments to be consistent with naming conventions of other safety charts. Keep support of previous argument names until next major release. This includes the following changes:

id -> id_col
major -> major_col
minor -> minor_col
group -> group_col
placeholderFlag -> missingValues
totalCol -> showTotalCol
diffCol -> showDiffCol
prefTerms -> showPrefTerms

See #3 for related request.

Not a problem with creating json via d3.csv() since everything is coerced to character, but can cause problems in other scenarios.

Fix is easy, just coerce the value to character here and here. This: return filter_d.currentValues.indexOf(rowData[filter_d.value_col])>-1 becomes this: return filter_d.currentValues.indexOf(""+rowData[filter_d.value_col])>-1

Change for "Serious" and "Open Query"

"All" at the bottom of the chart is rendering a page when it shouldn't be - this is breaking the system

Basically what is needed here is a way to prominently see and sort on SAEs (not simply embedded within AEs)
o Simple line listing would usually do
o Specific instruction to include SAEs as well as AEs

Also good would be to have a way (on the graphics site) to see which were SAEs
• line listing of SAEs (i.e., not necessarily differentiating AEs from SAEs within AE tables or graphs)
• some kind of highlighting of SAEs within the graph.
• being able to see SAEs graphically along with the others and a way of filtering to see just the SAEs

For the search bar and incidence threshold, the user must currently click away from the entered value or hit the enter button for the function to filter on the entered value. We should try to automate this so the user doesn't think the function is broken.

Issue: Selecting a group variable with many unique categories is likely to cause the browser to become unresponsive.

Solution: Add a check that halts execution if the user has selected a group variable with N>maxGroups categories. Could consider letting maxGroups be a parameter in the system with default=10 or so?

Confirm basic functionality and new features are working as expected in examples shown below:

Examples using v3.0.0 Release Candidate
Default

Basic Functionality (Tested every release)

• Check all spelling and grammar on the page
• View confidence intervals and tooltips for between-group differences by hovering the mouse over the - Difference Between icons.
• Expand the dataset to show nested preferred terms by clicking the "+" to the left of each category.
• Search by terms of interest and then close the search.
• Confirm that filters on events (e.g. Severity) are working as expected: rates in the table change, but number of observation in the table header do not.
• Confirm that filters on participants (e.g. Gender) are working as expected: rates in the table and number of observations in the table header both change.
• Select events based on a minimum incidence threshold
• Confirm that toggling between 'participant' and 'event' views work as expected (there should be more events than participants.
• Confirm that you can view a detailed listing of events by clicking on the title of a given row in the table, and that the current filters are applied as expected in the detail view.
• Confirm that the button that closes the detailed listing works as expected.

New Functionality (Tested in V3.0.0 only)

• (#4) Confirm that you can filter on participant demographic information (sex, median age, and treatment group when data is unblinded)
• (#5) Confirm that when a series of events is selected, the ability to filter on this detailed view is working as expected
• (#27) Confirm that group columns are sorted alphabetically
• (#32) Confirm that selecting a group variable with multiple categories doesn't crash the browser
• (#43) Confirm that the event counts in the column headers are updating when changes are made to the participant filters

Current implementation of mergedConfig in createChart.js in v3.0.0 is arguably better than nothing, but it's very naive and likely to cause bugs if the user specifies incomplete settings. For example, specifying settings.plotSettings.w but not settings.plotSettings.h is likely to throw an error.

Let the user download the table data when chart.config.validation == true. An early version of the functionality is already present in aetable.init() and \util, but isn't documented or exposed to the user.

Remove all dependencies except for D3.

Be able to filter (e.g. sex, severity, etc.) in details view

...like settings.plot_settings.diffMargin affects difference chart.