Előadás címe: Evaluation of usefulness of the receptorial responsiveness method (RRM), a procedure based on nonlinear regression of biological data
Időpont: 2020.06.30., 09:40 – 10:00
Kivonat: In all previous works, RRM was carried out with assumptions of Gaussian distribution for the scatter of data points around the curve, and of the same extent of this scatter along the curve. However, the only reason to do this was the long-established observation that these assumptions are usually true for non-transformed (at most normalized, in certain cases logarithmic) data obtained from biological systems. Thus, one goal of our investigations was to explore whether other curve fitting options based on other assumptions ameliorate the accuracy and/or precision (and thereby the usefulness and/or reliability) of RRM. In addition, another goal was to compare two ways of curve fitting, the individual and the global one. While individual fitting is the conventional form of regression (as many E/c curves, as many curve fittings), the global fitting (simultaneous fitting to all related E/c curves) is thought to be a powerful procedure that can reduce the uncertainty experienced when the related E/c curves are individually fitted. We have found that the best estimates of RRM can be obtained via individual fitting without any weighting, almost irrespectively of the fact of whether ordinary (assuming Gaussian distribution) or robust (optimized for Lorentzian distribution) regression is chosen. However, regarding the reliability, it is worthwhile to perform an ordinary regression, as only this method provides 95% confidence intervals for the best-fit values informing us about precision of the curve fitting. Accordingly, RRM is a relatively (by comparison to the possibilities it offers) easy-to-use procedure that requires neither a heavy-duty curve fitting software nor a high level of knowledge concerning regression analysis.