Module 05 / safety signal review
Safety Signal Review
The four views a safety reviewer reaches for first when a study unblinds or a signal is suspected: hepatic safety against Hy's law, time to the first moderate-or-severe event, adverse event (AE) incidence by system organ class with risk differences, and a baseline-to-worst lab shift. Nothing here is pre-rendered — the Kaplan-Meier steps, confidence intervals and shift counts are computed from record-level synthetic data in your browser as you read this.
240 synthetic subjects, two arms. Any apparent imbalance is a property of the seeded random generator, not of a real compound.
Hepatic safety — peak ALT vs. peak total bilirubin
Each point is one subject's worst on-treatment value, as multiples of the upper limit of normal (log₂ axes). Dashed lines mark 3×ULN (ALT) and 2×ULN (bilirubin).
Time to first moderate or severe treatment-emergent AE
Kaplan-Meier estimate by arm; ticks mark censored subjects. Computed from raw event times in the browser.
Swipe horizontally to see all columns →
AE incidence by system organ class — risk difference
Investigational minus placebo, percentage points, with normal-approximation 95% CIs. Sorted by risk difference. Unadjusted for multiplicity — a screening view, not a confirmatory test.
Lab shift — baseline vs. worst post-baseline
Swipe horizontally to see all columns →
Cell color intensity scales with the subject count; cells above the diagonal (shift to a worse category) are tinted rose.
All subjects, events, and laboratory values on this page are synthetic, generated by a seeded script (tools/platform-demo/generate-data.mjs) and embedded at build time. No patient data, no network calls.