1887
banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Choosing the optimal fit function: Comparison of the Akaike information criterion and the F-test
Rent:
Rent this article for
USD
10.1118/1.2794176
/content/aapm/journal/medphys/34/11/10.1118/1.2794176
http://aip.metastore.ingenta.com/content/aapm/journal/medphys/34/11/10.1118/1.2794176

Figures

Image of FIG. 1.
FIG. 1.

Scheme of the F-test selection process. Note that depending on the data different combinations have to be tested [in this case (Table IV, P5) the combination was necessary].

Image of FIG. 2.
FIG. 2.

Time activity curves for blood together with the best fit curves (Tables III and IV) for both model selection criteria.

Tables

Generic image for table
TABLE I.

Regression diagnostics of the best model (for and no weighting) selected using AICc.

Generic image for table
TABLE II.

Regression diagnostics of the best model (for and no weighting) selected using the F-test.

Generic image for table
TABLE III.

Akaike weights for weighting and no weighting. (Note: The Akaike weights indicate the probability that the model is the best among the whole set of considered models. The highest weight, i.e., the best model, is presented in bold.)

Generic image for table
TABLE IV.

values for weighting with and no weighting. [Note: The F-test compares the models and rejects the simpler model if the value is below the significant level (here ). It is not sufficient to compare models most similar in complexity (Fig. 2). In two cases it was not necessary (NN) to state a value, because the increase in model complexity (more parameters) did not improve the fitting.]

Generic image for table
TABLE V.

Frequency of selected models with AICc applying the Jackknife method. (Note: Nine times one data point was deleted and the AICc was determined with the remaining eight points. The best function with the original data is in bold. The AICc value for the “” function could not be calculated, because of too few data points.)

Generic image for table
TABLE VI.

Frequency of selected models with F-test applying the Jackknife method. (Note: Nine times one data point was deleted and the F-test was determined with the remaining 8 points. The best function with the original data is in bold.)

Generic image for table
TABLE VII.

Mean values of the pharmacokinetic parameters (11 patients). (Note: “Shared” means that these parameters were calculated for all patients together. “All nonshared” correspondingly means that the four parameters are calculated for each patient separately. All other combinations of shared and nonshared parameters have Akaike weights smaller than . The given standard deviations of the parameters represent the variations between patients and not the uncertainties due to the fitting procedure. The standard deviation of the Akaike weights reflects the different results deduced by the Jackknife crossvalidation, which is based on leaving one patient out 11 times.)

Loading

Article metrics loading...

/content/aapm/journal/medphys/34/11/10.1118/1.2794176
2007-10-18
2014-04-17
Loading

Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Choosing the optimal fit function: Comparison of the Akaike information criterion and the F-test
http://aip.metastore.ingenta.com/content/aapm/journal/medphys/34/11/10.1118/1.2794176
10.1118/1.2794176
SEARCH_EXPAND_ITEM