Chapter 5: Spread

As described in section 2.4 the Scenario parameter spread may be set to true resulting in random variations of several parameters. Handling spreading is illlustrated in figure 4.

Figure 4: Spread used by Simrisc components

When an analysis starts Loop::initialize passes Scenario::spread on to the member Loop::spread. Various components support random fluctuations of their parameters, and if Scenario::spread is true these random fluctiations are applied to these components. Figure 4 illustrates where random fluctuations may be applied:

the Mammo modality's systematicError;
the Screening's attendanceRate;
various Tumor components:
- Survival;
- Growth;
- TumorInfo:
  - Beir7;
  - Incidence's carrier specifications.

Random fluctuations can be applied to means, standard deviations and proportions.

For means the normal distribution is used by default. For example, the specification of the meanAge distribution, specifified for carrier Normal in the Incidence component of the Tumor specification may look like this:


    #                   value   spread  distr.                    
    meanAge:            72.9    .552    Normal

95% of the mean-values which are randomly drawn from this distribution are located in the interval 71.8 to 74.0. When spreading is used, thus computed mean values are used as mean values of the Normal carrier during the actual simulations.

For standard deviations the chi² distribution is used, ensuring that the standard deviation that is used for the simulations exceeds 0 (see the next section for a description of the theory and implementation in simrisc). For example, the standard deviations of the Normal carrier distribution may look like this:


    #                   value   spread  distr.                    
    stdDev:             21.1    .048    Normal

The number of degrees of freedom of the chi² distribution to use N - 1, where N is the number of cases that are used in a simulation.