LoginSign Up
← Back to all sparks
D

DoseFinding

ANALYTICS
Velocity0.0

DoseFinding — R package from the openpharma project

github.com/openpharma/DoseFinding ↗

New stewardship at openpharma, then two releases adding the methods MCP-Mod was missing

dose-responsemcp-modclinical-trialsmodel-averaginginterim-analysisopenpharma
◆Current state
DoseFinding implements MCP-Mod and related dose-response methodology for clinical trial design and analysis. In 2024 it changed hands — Marius Thomas took over as maintainer, Novartis was recorded as copyright holder and funder, and the package moved to the openpharma GitHub organisation with roxygen documentation and a proper NEWS file. The two releases since have added substantive methodology: model averaging for dose-response fitting in 1.3-1, and conditional and predictive power for interim analyses in 1.4-1.
◆Where it's heading
The pattern before the handover was maintenance — R-devel compliance, a bug fix, a link. After it, each release carries a named methodological addition with an acknowledged contributor, plus documentation to match: a longitudinal analysis vignette shipped alongside the interim power work. Housekeeping continues underneath, mostly clearing deprecated ggplot2 interfaces, aes_string in one release and qplot in the next.
◆Prediction
Given the last two releases each added one method with a supporting vignette, expect the next to follow the same shape. Both additions so far extend the package beyond fixed dose-response fitting, so adaptive and interim methodology is the more likely direction.

◆Recent moves

  1. 1y ago

    Conditional and predictive power for interim analyses

    Adds powMCTInterim for conditional and predictive power at an interim analysis, extending the package from designing a trial to monitoring one in progress. A vignette on analysing longitudinal data ships alongside it, and deprecated ggplot2::qplot calls are replaced.

    View source ↗
  2. 1y ago

    Model averaging arrives for dose-response fitting

    ⚡ SPARK

    Adds maFitMod, which fits dose-response models by averaging across candidate models rather than selecting one, and corrects ED dose estimation with type = "discrete" — the maximum effect was previously taken across the doses passed as arguments rather than across the fitted model's observed dose range, so existing estimates change. The first methodological addition after the openpharma handover.

    View source ↗
  3. 1y ago

    Package moves to openpharma under new maintainership

    The stewardship release: Marius Thomas becomes maintainer, Novartis Pharma AG is recorded as copyright holder and funder, and the package moves to the openpharma GitHub organisation. Documentation converts to roxygen with a README and NEWS replacing the old changelog, and former Depends packages become Imports. No methodology changes, but every release after this one does add methodology.

    View source ↗
  4. 2y ago

    Compliance update for R-devel

    A compliance update for the then-current R-devel. Representative of the pre-handover cadence, where releases responded to the toolchain rather than adding anything.

    View source ↗
  5. 3y ago

    powMCTBinCount bug fix

    Fixes a bug in powMCTBinCount and adds a link to the DesignMCPModApp. A two-item patch release.

    View source ↗