GIR Smoothing Simulator
The standard smoothing procedure used during the analysis of glucose clamps for the evaluation of glucose infusion rates is LOESS. LOESS not only helps with the graphical interpretation, but also with the calculation of parameters like cmax and tmax.
The only parameter of LOESS that is usually published is the smoothing factor. The smoothing factor is usually in the range of 0.1 for short acting insulin and around 0.25 for long acting insulin. By changing the smoothing factor, insulin curves can get flatter (e.g. larger smoothing factor for long acting insulin) or peaks more pronounced (e.g. smaller smoothing factor for short acting insulin).
But there are more options to influence the outcome of the LOESS smoothing: different implementations in different statistical programs can lead to different outcomes. In addition different versions of LOESS in the same program can also produce a different results.
The form on the right opens a simulation for different insulin types, different smoothing factors and different LOESS implementations. You can choose between
LOESS which can be regarded as the “standard” LOESS
LOESS with a linear weight function, which produces a flatter profile
LOWESS which is a mixture of the above using a different implementation approach