mfp2 - Multivariable Fractional Polynomial Models with Extensions

Multivariable fractional polynomial algorithm simultaneously selects variables and functional forms in both generalized linear models and Cox proportional hazard models. Key references are Royston and Altman (1994) <doi:10.2307/2986270> and Royston and Sauerbrei (2008, ISBN:978-0-470-02842-1). In addition, the implementation can model semi-continuous covariates with a “spike at zero” using a two-stage selection procedure. This extension follows the framework proposed by Becher et al. (2012) <doi:10.1002/bimj.201100263>. The package also includes the approximate cumulative distribution (ACD) transformation to model a sigmoid relationship between variable x and an outcome variable y, as described in Royston (2014) <doi:10.1177/1536867X1401400206> and Royston and Sauerbrei (2016) <doi: 10.1177/1536867X1601600>. This feature distinguishes it from a standard fractional polynomial function, which lacks the ability to achieve such modeling.