Estimation of non-parametric regression in the analysis of the anti-inflammatory activity of diverse extracts of Sideritis foetens

SUMMARY A procedure to choose the best non-parametric estimator from among all nonparametric methods to fit regression curves is described. The methodology that is proposed prevents a lack of fit at the edges of the regression curve. The method is summed up in a few steps to facilitate its application by researchers. The procedure is applied to the determination of various curves that explain the anti-inflammatory activity of diverse extracts of Sideritis foetens and phenylbutazone against the time elapsed from the application of the agent which provokes the inflammation. Discussion shows that it is possible to obtain valid conclusions about the effects of the different products and to establish comparisons between them. Such conclusions are not possible when starting from the classical statistics methods usually employed in pharmacology.