Semi-parametric ordinal regression models for continuous scales

We extend the ordinal regression model for continuous scales by introducing non-parametric terms in the linear predictor. The model parameters are estimated using constrained optimisation of the penalised likelihood and the penalty parameters are automatically selected via maximisation of their marginal likelihood. An application of the model in a study on alternative chemotherapy treatments for advanced breast cancer is shown. The outcome of interest, the quality of life of patients, is modelled with fixed effects, smoothing terms and individual random effects and its association with the treatment arm over the chemotherapy cycle is estimated. The methodology has been implemented in an R package, ordinalCont, available on CRAN.