Figure 1

PopKinetics analyzes populations of individual subjects, where each subject in the population is represented by one SAAM II Compartmental Study file. PopKinetics is compatible with only SAAM II v2.0 and is configured to system requirements for Microsoft Windows 95/98/Me; Windows NT 3.51 or later, or Windows 2000/XP/XP Pro, or Windows Vista.

 

General

• Standard Two-Stage (STS) and Iterated Two-Stage (ITS) parametric population analysis methods

• Operates directly on SAAM II single-subject models

• Full SAAM II functionality in population setting

• Allows population analysis of complex (multi-compartment) models; simple models analyzed quickly and easily

• Fast Setup (Figure 1) for population analysis (point and click)

• No programming or pseudo code required

• Setup independent of model complexity

• Analysis setup easily modified

• Experimental protocol can vary between subjects

• Incorporates a Global Tool to rapidly modify model and data attributes for each subject in the population

  • The Reference File provides a mechanism to modify the settings in every Subject File in a population without having to make the changes manually to each file. In other words, certain settings in the analysis files generated are inherited automatically from the Reference File. To change the data weighting from relative to absolute for the analysis, for example, simply change the setting in the Reference File and re-save the file. During the next Compute (or Check), PopKinetics automatically includes the new value(s) in each file that it generates from the originals in the new analysis.   Note that the original files are not modified.

• Requires few assumptions to begin population analyses

• User’s notes can be included in Analysis Files to document particular items of interest in an analysis: setup, conclusions, etc.

• Includes detailed on-line help with examples

^Top

Computations

• Computations proceeds automatically in populations with constant, fixed, and random effects

• Alerts users when there are problems in analyzing data; provides interactive error control; direct integration with SAAM II allows for speedy data and model error identification and correction

• Confidence Intervals (Figure 2) on population parameters to indicate precision with which population parameters are estimated. There is no mechanism in the Standard Two-Stage, Iterated Two-Stage, or Bayesian parameter computations that can compute confidence intervals directly. The confidence interval for each parameter is computed after a completed analysis by generating a number of simulated populations and selecting appropriate values. Specifically, Confidence Intervals are determined by discarding the highest and lowest parameter values from the computed values for the populations. Sufficient populations must be generated to allow at least one highest and one lowest parameter value to be discarded. To compute the 90% confidence interval, for example, requires a minimum of 20 populations. After the highest and lowest values are discarded, 18 of 20 population values remain, giving (18 / 20) * 100 = 90%.
  

Figure 2

 

• Progress Display during analysis (Figure 3)

  During an ITS or STS analysis, files are constantly being opened, "fitted," and re-saved. During an STS analysis, there is only one cycle. In an ITS analysis, there may be numerous cycles. As the computations proceed, the Analysis Progress window is constantly updated. In the upper pane, in the first column, the status of each file is indicated as the analysis progresses. A blue arrow indicates the file being processed. A green check indicates that the file was processed successfully. A red "X" indicates that there was an error detected while processing the file. The second column shows the name of the files. The third column shows the "cycle number." If "convergence" has been achieved, the value is shown in the last column (not used in an STS analysis).

  Figure 3

^Top

Output/Results

• Analysis results may be plotted using General plot or Samples (Figure 4) and Data plot (Figure 5) is available to examine frequently used results) or listed in Tables; Results (Figure 6)may be exported to other applications or printed to a file for additional analysis or reporting - flexible text reporting built in

• Provides detailed Logs  (Figure 7) )to examine events that occur during an analysis to allow user to verify that the analysis proceeded as intended

Figure 4		Figure 5
Figure 6		Figure 7

^Top

Powerful Simulation Capabilities

• Simulates Populations with variability in data and model parameters.

  • The PopKinetics simulator produces a simulated population based on a SAAM II Compartmental model. It can generate any number of Subject Files containing Random, Fixed and Constant Effects. PopKinetics uses the settings and values specified in the simulator window and the Reference File from the main PopKinetics window as the Base File. The simulator can generate a population containing variability in the parameters, noise in the data, or both. This provides great flexibility in determining the characteristics of the population.

• Simulates Clinical Trials to determine effect of varying dosing regimens

• Performs Monte Carlo Simulations that can be used to test single-subject models. Monte Carlo simulation can be used to conduct sensitivity analysis on variables and calculate better confidence intervals for an individual analysis than those available from asymptotic statistics like those available from the Statistics Window in SAAM II. Suppose one wonders if the statistics are adequate and the confidence intervals for the parameters are well calculated. Monte Carlo simulation allows us to test such a hypothesis, provided the statistics of the measurement error are known. It is based on simulating a large number of synthetic data sets and calculating the average estimate for all the synthetic data sets.

  •  If we wish to test the parameter confidence intervals, we are interested only in the effect of the measurement error.

  • We do not allow the parameters to vary (there is no between-individual variation).

  • The PopKinetics Simulator screen would appear similar to the screen in Figure 8.

  • Since we’re treating the parameters as constants for this simulation, the warning in Figure 9 will appear: This is “OK” for a Monte Carlo simulation. Once the simulated population has been generated, a quick STS analysis provides statistics that can be compared to the original SAAM II study statistics.

Figure 8	Figure 9

^Top