How we know
Every link on this study, and who decided it
The study on the left, the conditions and drugs this site links it to, and on each link the judgments that made or checked it, by day and by judge. The top row is the record itself: each new version seen since tracking began. Dashed lines are links the registry record does not make itself. Select a link to see everything judged on it, or a mark for one decision with its reason and evidence.
Times are in your time zone; a day marked “day only” kept no time. The graph scrolls sideways on a narrow screen; every mark is also in the list below.
Ipatasertib
From the registry record3 judgments on this linkAll judgments on this study
| When | Link | Judge | Decision | State |
|---|---|---|---|---|
| Oct 2 | The record itself | ClinicalTrials.gov record | Version of 2026-10-02 | Record change |
| Nov 3 | Ipatasertib | Claude Opus 5.5 | Tested | Replaced |
| Nov 3 | Tiragolumab | Claude Opus 5.5 | Tested | Replaced |
| Nov 3 | Atezolizumab | Claude Opus 5.5 | Tested | Read an older version |
| Nov 3 | Bevacizumab | Claude Opus 5.5 | Background | Read an older version |
| Nov 3 | Entrectinib | Claude Opus 5.5 | Background | Replaced |
| Nov 3 | Inavolisib | Claude Opus 5.5 | Background | Replaced |
| Nov 3 | Divarasib | Claude Opus 5.5 | Background | Read an older version |
| Nov 3 | Inavolisib | Claude Opus 5.5 | Tested | Read an older version |
| Nov 6 | The record itself | ClinicalTrials.gov record | Version of 2026-11-06 | Record change |
| Nov 6 | The record itself | Rule based · nightly check | Reading set aside | Current |
| Nov 10 | The record itself | Rule based · nightly check | Read again | Current |
| Nov 10 | Ipatasertib | Claude Opus 5.5 | Tested | Current |
| Nov 10 | Tiragolumab | Claude Opus 5.5 | Tested | Current |
| Nov 10 | Entrectinib | Claude Opus 5.5 | Background · care | Current |
| Nov 14 | Neoplasms | Rule based · condition-kind rule | General: a category, not a disease | Current |
| Nov 18 | Ipatasertib | Rule based · role rule | Tested | Agrees |
| Nov 18 | Tiragolumab | Rule based · role rule | Tested | Agrees |
| Nov 18 | Atezolizumab | Rule based · role rule | Background | Overridden |
| Nov 18 | Atezolizumab | Rule based · role rule | Tested | Current |
| Nov 18 | Bevacizumab | Rule based · role rule | Background | Current |
| Nov 18 | Entrectinib | Rule based · role rule | Background | Agrees |
| Nov 18 | Inavolisib | Rule based · role rule | Background | Overridden |
| Nov 18 | Divarasib | Rule based · role rule | Background | Current |
| Nov 18 | Giredestrant | Rule based · role rule | Background | Current |
| As of now | Neoplasms | Rule based · main-condition rule | Main condition | Current |
The rules, in plain words
“Rule based” means one of these decided, not a model or a person. Each runs every night over every study.
Which drugs a study tests
Reads each treatment the study lists and the arms it is in. A drug only in the experimental arm is tested. A drug in every arm while something else changes between the arms is given beside it. A drug only in the control arm is the comparator. Placebo, procedures and usual care are known by name.
Where the whole record has been read, that reading's answer replaces the rule's.
What a study is mainly about
Starts from the first condition ClinicalTrials.gov files the study under. It moves to a narrower one only when the study's own title or listed conditions name it, and sets aside terms the registry often files by mistake. A study open to many cancers is not placed under one of them unless its title names it.
Checked against studies judged one by one; the study page says when it differs from the registry.
How a condition link is counted
For each condition a study is filed under: whether it is the study's own condition at the finest the study states it, a wider one, a general category, or filed by mistake; and whether it is the main condition, a setting the study names, or only how it was filed.
A decision on file for one study overrides it, with who made it and when.
When a study changes after it was read
Every night, studies whose record changed are compared with the last reading. A change to sites or dates leaves the reading as it is; a reworded entry keeps it; a change to the treatments or conditions marks it to be read again.
Until it is read again, the page shows the last reading and says so.