The purpose of performing Monte Carlo simulations in kinetic modeling is to get information about the standard error of the parameter values found in a fit. There are two different kinds of sources which contribute to the uncertainty of the parameters:
The paradigm used for performing Monte Carlo simulations is described by Flannery et al. . The basic idea is to simulate a series of measurements and statistically analyze the results when modeling these synthetic "measurements".
In PKIN Monte Carlo simulations can be performed for compartment models applied to a single tissue and models coupled across several tissues by selecting the respective option button on the Mt-Carlo panel.
Three types of "noise" can be defined. For each "noise" the distribution type as well as the variance has to be configured. After selecting the distribution a dialog window appears for defining the variance as described in the Residual weighting for model fitting.
Three different distributions are available for noise generation, Gaussian, Poisson, and Uniform. They are applied for each individual measurement with the prescribed variance.
For the Model noise simulations there is also the Bootstrap method available. With the Bootstrap approach, the weighted residuals wri of the initial fit are used . For simulating a noisy curve, these residuals are randomly selected and added to the model curve. Specifically, the noise for a frame j is constructed from a randomly selected wri of frame i by dividing it by the weight of frame j. As a consequence, it will have the same effect in the fit as it originally had in frame i.
When the Start Monte Carlo for current region button is selected the following processing starts:
It is possible to visualize a summary of the fit results during the Monte Carlo runs by switching back from the Tissue to the Mt-Carlo pane. In the curve display, a new curve Monte Carlo appears with vertical bars. The end points of these bars mark the minimal and the maximal value that any of the resulting model curves reached at a particular time point.
The value of using the fitting history is illustrated below after a Monte Carlo simulation with enabled Add results to fitting history. Each entry from the simulation has a common Revision number. The Monte Carlo Initial Fit line contains the result of the first fit to the original data. If Add noise to measured TAC is not selected, the model curve is the curve to which the generated noise is added. The following entries Monte Carlo Simulation are the individual fits to the generated noisy TACs. The final line Monte Carlo Result represents distribution of all individual fits: the parameters are calculated as the average across the individual fits, and the standard error as their standard deviation.
Double-click into the history to show the noisy TAC and the fit result of a simulation. Use the arrows
to conveniently step through the simulations.
Note that the results of the simulations can also be visually presented by selecting all simulations in the Parameters history and selecting the other panels. The Display parameters and Display relative change curve show the individual parameters across the simulations as absolute and relative values, respectively. Display mode curves shows all noisy TACs with the respective fitted model curves.
The individual fit results can also be saved for dedicated statistics. To this end select an entry with the Revision of the Monte Carlo simulation, then use Save Selected Revision as illustrated below.
Note: When saving a .km file, the history is included. Therefore the file created can become large when including extended Monte Carlo simulations.
Several Tissues Coupled
The same principle as for the single tissue Monte Carlo simulation applies to a coupled Monte Carlo simulation. It can be started by Start Monte Carlo for current region and uses the configuration prepared on the coupled panel. Model noise uses the weighting individually defined for each coupled region. The fitting history of all coupled regions uses the same Revision number, so that all results can be exported at once.