NEWS

---

PSPI 1.4

• Add support for persistent model fitting (PSPI_fit(), PSPI_predict()).
• SplineBART and DSplineBART now resolve the BART/spline non-identifiability by dropping the spline intercept instead of centering BART; the previous centered implementations are removed.
• "DSplineBART" is accepted as a model name; "MSplineBART" remains an alias.
• sim_trans() and sim_generalizability(): participation model of the nonlinear scenario updated.
• Title updated to cover transportability.

PSPI 1.3

Add support for binary outcomes.

PSPI 1.2 (2025-12-02)

Fix the bug in the initialization of BART.

PSPI 1.1 (2025-12-01)

This release introduces the PSPI package, which implements Propensity Score Predictive Inference methods for generalizing trial findings to a target population.

Key Functionalities:

• PSPI_generalizability: Use Propensity Score Predictive Inference (PSPI) for generalizability analysis.
• sim_data: Simulate data for trial data and population data.

License:

This package is licensed under the GNU General Public License version 2 (GPL-2). The core computations leverage code derived from the BART3 package, originally developed by Rodney Sparapani, which is also under GPL-2.

Known Issues:

• No significant issues at the moment. For more details on potential limitations, please refer to the documentation.