ISoP ISoP on Twitter | ACoP ACoP on Twitter | PAGE | WCoP | PAGANZ | PAGJa

Less than 3 model selection criteria (MSC) and 0.89-0.93 of coefficient of determination~ bad fit?


Dear all,

I just completed nonlinear regression curve fitting of an inhaled drug in healthy volunteers. The drug follows a two-compartment open body model and I can see that the curve fitting yields a decent fit through the observed data points (visually). When I look at the MSC and COD though, they are fairly low (MSC= 2.8 and COD 0.89). Also the 95% CI of the parameter derived was fairly large (3 fold). So I am wondering if this is considered a bad fit and what could be the reasons for it? However, interestingly when I look at the curve fitted profile, it seems to go through most of the data points. So how does one address this, and why is this usually seen? I didn’t use any weighting factor when curve fitting because it didn’t make a difference to the goodness of fit visually, and further compromised the MSC and COD. Thank you.