A visual guide to one global question from many repeated outcomes.
They give us a grid of measurements: several clinically important outcomes, measured repeatedly over time.
“Which single box is significant?”
The treatment effect may be spread across the whole pattern rather than concentrated in one dramatic measurement.
“Across the outcomes and visits taken together, does treatment tend to do better overall?”
Think in terms of who tends to rank better.
LRST is looking for a repeated overall advantage — not one dramatic box.
That makes the method natural when the scientific question is about overall treatment benefit across multiple repeatedly measured outcomes.
“Is there an overall treatment advantage across important outcomes and visits?”
This is the sweet spot for LRST: several outcomes, repeated visits, different scales, and a desire for one robust global test rather than many scattered p-values.
“Exactly how much does this specific endpoint change at this specific time?”
If the real goal is a precise mean difference, an individual trajectory, a treatment-by-time interaction, or prediction, then a more targeted longitudinal model is usually the better fit.
The same research program includes extensions for increasingly complex longitudinal settings.
This workflow follows the current package README and exported R functions.
No wrapper is needed: the documented package API works directly with the LRST functions.
For the two-arm test, provide one placebo array and one treatment array.
The returned object includes the test statistic, standard error, p-value, visit-level score vector, covariance estimate, and relative treatment-effect estimates.
The package estimates power at the chosen significance level using the observed placebo and treatment arrays.
The documented multi-arm function returns the p-value, selected arm index v, and final test statistic T.
Package workflow based on the current djghosh1123/LRST GitHub README and exported function documentation.