Abstract
A general approach to semiparametric regression model construction is proposed, using the notation of a transfer functional. Estimators of unknown regression parameters are proposed, generalising the estimators of the parameters of the proportional hazards and the additive risk models.