We propose a semiparametric mixed–effects model (SNMM) using penalized splines to clas- sify longitudinal data and improve the prediction of a binary outcome. The work is motivated by a study in which different hormone ...