LoginSign Up
← Back to all sparks
J

JointFPM

INFRA · APIS
Velocity0.0

A Parametric Model for Estimating the Mean Number of Events

github.com/entjos/jointfpm ↗

Recurrent-event modelling settles, with mean_no() promoted to stable.

survival-analysisrecurrent-eventsparametric-modelsapi-stabilityprediction
◆Current state
JointFPM fits joint flexible parametric models for a recurrent event process alongside a competing terminal event, and predicts the mean number of events. The visible history runs from bug fixes on the earliest CRAN releases through standardization, integration options and a summary method, ending with mean_no() declared stable. Several changes arrived through outside pull requests.
◆Where it's heading
The arc runs from a working estimator toward a usable one: input validation and error messages first, then control over the numerical integration, then a summary method and pass-through arguments to the underlying rstpm2 fit. The latest release adds no code so much as a stability commitment to a function users were already calling.
◆Prediction
With mean_no() stable, the next work most likely targets the prediction and standardization paths rather than the model fit itself.

◆Recent moves

  1. 1y ago

    mean_no() promoted to a stable interface

    A one-line release that moves mean_no() out of provisional status. Nothing else changes, but downstream code can now depend on its interface.

    View source ↗
  2. 2y ago

    summary() method and control arguments passed to rstpm2

    Adds summary.JointFPM() for reading model estimates, and control and dots arguments on predict.JointFPM() that pass through to the underlying rstpm2 fit, which is what users reach for when estimation needs tuning.

    View source ↗
  3. 2y ago

    Gaussian quadrature option for the mean-events integration

    Gaussian quadrature becomes an alternative to Romberg integration when computing the mean number of events, trading node count for speed. It matters most when standardizing over continuous covariates.

    View source ↗
  4. 2y ago

    Standardized marginal estimates plus input validation

    Adds standardization so the mean number of events and differences between groups can be reported marginally, alongside input checks and clearer errors. This is the release that made the package's estimates reportable rather than merely computable.

    View source ↗
  5. 2y ago

    Bug fixes for differences between mean-event functions

    Corrective work days after the first release, on the estimation of differences between two mean-event functions. The fixes change reported numbers rather than internals.

    View source ↗