Indirect Estimation and Simulation Tool (IEST) Populations

11. Populations

A population (also referred to as population set or data set) represents a pool of subjects and their characteristics. A populations can be either input as data (to be used in a Simulation or an Estimation), or by specifying a distribution (to be used in Estimation or for randomly generating population sets).

11.1 Creating Populations

1. From the main form, click the 'Populations' button on the left-hand navigation pane. Note that this form can also be accessed by drilling down from the project form.

2. This form shows the population groups. Click the 'Add' button (A), and a new row will appear.
3. Enter the name for the population set in the box (C).
4. Click the 'Data' button (E) to define the population characteristics and associated data/distributions. The following form will appear.

5. To input a population as data: Add a characteristic by selecting a parameter from the table in the lower left (D) and Click the up arrow (C). To remove a row, highlight it, then click the down arrow (B). After all the population characteristics have been added, press the Data Tab and fill in data for the chosen parameters for each individual. The data can also be imported from a file using the Import button. To view or change the data, press the data tab (G).
6. To specify a population by its distribution: Add a characteristic by selecting a parameter from the table in the lower left (D). Additionally, define the distribution expression in the text box (E), or select a distribution from the table in the lower right (F) and it will appear in (E). Click the up arrow (B) to add a row that combines the distribution and the parameter. To remove a row, highlight it, then click the down arrow (C).

NOTE: When attempting to duplicate a population defined by characteristics published in the literature, it is essential to recognize that the characteristics are not independent. For example, systolic and diastolic blood pressure are very highly correlated (r = ~0.8) and height and weight are also correlated (r ~ 0.5). For example, if height and weight are generated independently, it is likely that some subjects will have extreme body mass indices (BMIs) (i.e., a low weight in a very tall person may yield a physiologically unrealistic low BMI and conversely). Correlated covariates can be generated within the system, by first generating one covariate and then making the second covariate depend on the result of the first (i.e., express the second covariate by a regression function that is dependent on the first covariate).

For a video demonstration on creating a population, click here.

11.2 Removing an Entire Population

Identify the population, and click the 'X' (delete) button. This may require deletion of other entities and may be difficult if the deletion candidate was extensively used.

11.3 Generating new population data based on distributions

The system supports the automatic generation of a population set defined by distributions of its characteristics. This feature can be used to automatically generate population sets in the system according to distributions provided in the literature. To perform these tasks, the following steps should be taken:

  1. Follow the steps defined above in Creating Populations to define a population set defined by distributions that were defined in the previous step.
  2. Select the desired distribution based population. The population set should read "Distribution based" in the Definition Type field (D).
  3. Right click the mouse and a pop up menu will appear. Select the entry "Regenerate New Population Data from a Distribution".
  4. A new input dialog window will appear and will ask for the population size to be generated. Enter the desired number and press OK.
  5. The system will generate a new population set filled with data that was generated according to the distributions defined in the originally selected distribution population set. The system will place this population set at the end of the list and will focus on it so the user can modify it.

Note that generation of a data based population from distributions is controlled by multiple system option parameters that are listed in Parameters.