QT Safety Explorer

gsm.safetyexamples

QT/QTc central tendency against the ICH E14 threshold, outlier scatter, and categorical summaries of participants crossing the 450/480/500 ms boundaries.

gsm.safety 1.0.0 · safety.viz 1.4.0 · rendered 2026-07-23

QT/QTc cardiac safety across three linked views. The chart below is live — rendered by the safety.viz qtExplorer module on the package’s example ECG data, with the full control sidebar.

Central tendency — mean change from baseline (Δ) by visit and arm, and, when placebo_arm is set, the placebo-corrected change (ΔΔ) with confidence intervals against the ICH E14 10 ms threshold of regulatory concern.
Outliers — a scatter of absolute values against change from baseline, with the 450 / 480 / 500 ms and 30 / 60 ms category boundaries drawn in.
Categorical — counts and rates of participants crossing each of those boundaries, by arm.

The bundled adeg dataset uses the ADaM ADEG column names the module expects (TEST, STRESN, BASE, ARM, plus VISIT/VISITNUM/ABLFL), so the chart initializes with no settings at all; the settings below name the placebo arm to unlock the ΔΔ view and add subgroup filters.

This is safety.viz QT Explorer Phase 1. QTcI, PR/QRS intervals, participant drill-down, and hysteresis plots are Phase 2 work, tracked upstream.

A note on the demo data. adeg comes from the CDISC Pilot 01 ECG data, which — like plenty of real trial data — is internally inconsistent: it collects RR and heart rate as separate measurements, and the two contradict each other (they were generated independently, so only 0.8% of readings agree). The QTc values here are therefore cleaned at ingestion rather than taken from the source’s own pre-derived parameters: QTcF = QT / (RR/1000)^(1/3) and QTcB = QT / (RR/1000)^(1/2), correcting against the RR implied by the recorded heart rate, which is the more credible of the two inputs. See safety.viz#79 for the full working.

Even so, the pilot’s measured QT intervals are long for a resting population (median 444 ms), so QTcF centres near 468 ms and a majority of participants cross the 450 / 480 / 500 ms categories below. No correction formula can address that — it is a property of the synthetic source, not of the chart. The views exercise every threshold, but the crossing rates here should not be read as clinically typical.

Render this report from a workflow

The same chart ships as a gsm-idiom report workflow. Run it with gsm.core::RunWorkflow() to produce a standalone HTML report, or use the matching runner script:

lWorkflow <- yaml::read_yaml(
  system.file("workflow", "3_reports", "qt_explorer.yaml", package = "gsm.safety")
)

gsm.core::RunWorkflow(
  lWorkflow = lWorkflow,
  lData = list(dfResults = ExampleData("adeg"))
)

Rscript inst/examples/qt-explorer.R <output-dir>

The full qt_explorer.yaml workflow

meta:
  Type: Report
  ID: qt_explorer
  Output: html
  Name: QT Safety Explorer Report
  Description: Interactive safety.viz QT/QTc safety explorer with central tendency against the ICH E14 threshold, outlier scatter, and categorical summaries.
  lSettings:
    placebo_arm: Placebo
    filters:
      - value_col: SEX
        label: Sex
      - value_col: RACE
        label: Race
      - value_col: SITE
        label: Site
spec:
  dfResults:
    USUBJID:
      type: character
    TEST:
      type: character
    STRESN:
      type: numeric
    BASE:
      type: numeric
    ARM:
      type: character
steps:
  - output: strOutputDir
    name: getwd
  - output: lWidget
    name: gsm.safety::Widget_QtExplorer
    params:
      dfResults: dfResults
      lSettings: lSettings
  - output: strReportPath
    name: gsm.safety::SaveWidgetReport
    params:
      widget: lWidget
      strOutputDir: strOutputDir
      strOutputFile: ID

The live chart

library(gsm.safety)

dfResults <- ExampleData("adeg")

Widget_QtExplorer(
  dfResults,
  lSettings = list(
    placebo_arm = "Placebo",
    filters = list(
      list(value_col = "SEX", label = "Sex"),
      list(value_col = "RACE", label = "Race"),
      list(value_col = "SITE", label = "Site")
    )
  )
)