samussiah/animated-swimmer-plot

animated-swimmer-plot's Issues

Base all dimensions on x- and y-scales.

rather than positioning elements manually with width/height/margin

Add an introductory, static explanatory screen - color coding, x-axis, y-axis - like a tutorial.

Add interactivity.

Click a bar to view the participant's details or to highlight that participant and their progression

Implement vertical grouping.

y-axis stratification: response

as in stratifying vertically rather than by color - color will be used to distinguish stacked bars

decision: single canvas or paneled canvases? probably single canvas in order to transition the bars between groups but I wonder if a single svg couldn't contain one g per group... Still, sounds difficult to transition between two different groups.

Refactor before v1.

Add a control to change participant characteristic stratification.

Display time of current response in that view.

Update ID attribute of bar groups with ID of participant.

Add stratification by participant characteristic.

this feature might involve paneling or some sort of horizontal aesthetic for each participant to indicate their characteristic - solid line, dotted line, etc. or background color/shade of each horizontal band

Bars should enter from their start position.

Figure out how to handle negative timepoints.

Add a sorting to mimic a survival curve.

perhaps draw the survival curve then have the bars transition to their full width - play with the animation a bit

Support simpler states.

OS/PFS rather than RECIST categories of CR/PR/SD/PD

Update axis to display weeks or months instead of days.

Add setting to turn participant IDs on and off.

Add a control to sort by current response or by time to current response.

Reverse the direction of the plots.

i.e. the origin of the plot on the left is on the right of the canvas

Add htmlwidgets wrapper for use in R.

Add speed control.

Improve shuffling mechanic.

transition one response category at a time or one direction at a time: responders then non-responders

Replace stacked bars with box plot in overall survival sort.

Allow for a dynamic x-domain.

i.e. the maximum would be the current timepoint of the animation

Implement custom events.

e.g. clinical events of interest. This feature probably involves a second data source...

Position participants initially in the category of their first response.

to avoid excessive reshuffling

Add visit-level pauses.

At median or maximum study day of each visit, cycle through views and highlight current response/survival/durability statistics.

Add support for Best Overall Response.

e.g. a flag that identifies the BOR record and an option to sort by BOR and display counts of BOR. Reason being is BOR has additional criteria like multiple measurements of a given response

Implement stacked bars.

bar stratification: response

Add player.

play/pause, step backward, step forward, restart, range slider

Start with a bar chart race.

x-axis: time
y-axis: participant
animation variable: time

question: the x-axis will effectively be a loading bar because time is the animation variable. How can we better use the x-axis?

Use canvas as instead of range slider to select timepoint.

Add view control.

overall survival, current response, progression-free survival

Sync speed of animation with rate of change.

e.g. speed up between visit windows and slow down around visit windows or give a shorter duration to timepoints without change and a longer duration for timepoints with change

Add durable response control.

Communicate durable response.

• via annotation of a vertical line in current response view or custom aesthetic in other views
• via a flag in the dataset

Add destroy method.

Add delay at beginning.

Implement resize.

SME input

• oncology is a broad therapeutic area - need to focus on sub-areas of oncology
• solid tumor - how quickly and how significantly you can shrink the tumor
• hematological - decrease/increase of parameter
• PRO data - how can it be incorporated alongside traditional clinical data - e.g. quality of life as a differentiator when response rate is only non-inferior
• spider/radar chart - choose a set of continuous/ordinal outcomes and animate change from baseline
• short term and long term dynamics - quality of life
• capture event or progression itself in the bar chart race -
• think about using the solar system animation to visualize state change
• radar chart for PRO? check out Doug's PRO data visualization
• composite PRO data against tumor response data in a Gapminder-esque animated scatter plot over time