Changing the Model

Basics of Raedfast modelling

 

In this section we will look at the basics of Raedfast modelling:

  • copying an existing step and modifying its rules

  • creating a new scenario, and linking the new step to it

  • processing the new scenario

  • viewing the output in Excel

You can watch the video below, or read through the section text which follows the order of events in the video and adds more detail.

It is recommended that you read section Viewing Raedfast Output as a preliminary.

The video below takes you through the basics of Raedfast modelling, making a simple change to the model to create a new scenario and compare it with existing scenarios.

Opening the Raedfast Model

 

Open the Raedfast front end application.

You will see three main headings: Models, Dimensions and Data Tables. The data tables are ActivityHistory and ActivityModel. The dimensions are the dimensions of these data tables.

There is a single model, called Activity. Double click to open, or use the Open button at the bottom of the screen.

The Steps and Scenarios screen

 

You now see the main screen for the model, known as the Steps and Scenarios screen. Down the left of the table are the steps defined in the model (these are the elements of the ActivityStep dimension). Across the top are the scenarios defined in the model (the elements of the ActivityScenario dimension). In the middle we define which steps are included in which scenarios.

Copying a step and making a change

 

Creating a new step

The simplest way to create a new step is to copy an existing one. Alternatively, click Insert to create a new, empty step.

Choosing a model function

If you create an empty step you will need to assign a function to it, by double-clicking on the ModelFunction column. Refer to the function catalogue for a description of the available functions. Since we copied an existing step, our new step inherits the function of the old: this is the Add function, which adds to or subtracts from a given body of patients.

Naming the step

The system gives the new step a default name. You can change this, but in this example we will leave the default of Non-Demographic Growth 1.

Order of steps

You can also move the step up or down the list of steps. The order in which steps are processed is significant, because (unless you specify otherwise) each step takes as its input data the sum of the preceding steps. So the impact of a step will depend on which steps precede it.

Now click on the Save button at the bottom of the screen, to save the new step.

Opening the new step and changing its rules

 

Rules

Open the new step, either by double-clicking on it or by clicking the Open button at the bottom of the screen.

The step opens in a new tab, showing the rules of the step, as they were copied from the old Non-Demographic Growth step.

There are three rules. Each applies to a different Pod (Point of Delivery). Associated with each Pod is a value in the Number column. This number is the percentage change to be applied to that Pod. For Day Cases, Inpatients and Outpatients the Number value is 0.5, indicating a 0.5% increase.

The Year column

The Year column indicates the years to which each rule applies. In this instant the year is a rollup of all the years in the modelling period. So the system will apply the 0.5% to each successive year of the model.

Identifying cohorts of patients

The body of patients to which a rule applies can be defined using any dimensions of the model, and the selection from each dimension can be at the dimension base level, or at a rollup level. In this instance, DC is a base-level Pod, while All Inpatients and All Outpatients are rollups.

So here we could make any number of changes. We could create any number of new rules, using different dimensions of the model (click on Show Blank Columns to show available unused dimensions), we could apply absolute rather than percentage numbers, and we could replace rather than add to the existing activity.

Changing the modelled percentage growth

But for this example we will simply change the Number field, for Day Cases and Inpatients, to 1.0.

Click the Save button, and return to the Steps and Scenarios screen.

Copying a scenario

 

Creating a new scenario

Now we will attach the new step to a new scenario.

We could create a new blank scenario by clicking Insert. But we want this new scenario to be the same as the Mid Range scenario, except that the new scenario will use the new step Non-Demographic Growth 1, instead of the old, so we will begin by copying the Mid Range scenario.

Order of scenarios

The system places the new scenario to the right of the scenario copied. We could move it left or right if we chose. The order in which scenarios are listed on the Steps and Scenarios screen is significant, because that is the order in which they are processed. This can be important if we are using the output from one scenario as the Baseline to another.

In this instance though no such consideration applies, and we can leave the scenario in its default position.

Linking the new step to the new scenario

 

Now for the new scenario, Scenario 1, click off the old step, and click on the new. Scenario 1 now comprises the same steps as the Mid Range scenario, except that it uses the new Non-Demographic Growth 1 step with its higher figures, instead of the old.

Processing the scenario

 

At the bottom of the screen, click the Process Model button.

The Process screen for the model opens in a new tab.

Setting processing parameters

Here we can select the scenarios we want to process, and the year to which we want to run. The other two scenarios in the model — Do Nothing and Mid Range — are already processed, so there is no need to process them again. Click these Scenarios off, and click the new Scenario on. Select Year 5 as the end of processing in the Stop at Period field.

The system displays the run list in the right-hand pane. The run list is the list of tasks the system will go through, in the order in which they will be prosecuted.

Order of processing

The system processes the model rules in the following order:

  • step, within year, within scenario

That is, the system begins with the first scenario marked for processing, the order being determined by the position of the scenario in the columns of the Steps and Scenarios screen. It then selects the first year for which the scenario has any applicable rules, provided that year is before the year set in the Stop at Period field. Finally the system takes the first step in that scenario which has rules which apply to that year. Processing continues, following this order, until the last step of the last year of the last scenario has been calculated.

Each item in the run list then is a combination of a scenario, a year and a step, with the exception of the start run and end run items.

Processing

Click the Process button.

Note that when a scenario is processed, all existing model output for that scenario is first deleted. So if a scenario were processed as far as year 10, and was then re-processed but only as far as year 5, Model output for years 6 to 10 will be deleted.

Stopping processing

You can stop model processing by clicking on the Stop button.

Processing will stop when the system has completed the run list item on which it is working. The processing done up to the point at which it was stopped is not rolled back: data generated in the ActivityModel table by run list items executed before the process was stopped remains in the table.

Viewing the output in Excel

 

Automatic update of Analysis Services

At the end of processing, the system automatically updates the ActivityModel table in the Analysis Services database.

Most of the work being done during the End Run item in the run list consist of this update.

So when processing is complete you can switch to Excel, refresh your pivot table display, and observe the impact on modelled bed numbers of the new step compared with the old.

Pro-rating of the impact of assumptions across the base of the model

Raedfast automatically drives the impact of every modelling assumption down to the bottom level of the model, irrespective of the level at which assumptions are entered.

So although the cohorts of patients in the Non-Demographic Growth Steps are defined in terms of Pod only, in the pivot table the impact of these rules can be analysed at any level, across all the dimensions of the model.

This is important, as it enables each step to build cumulatively on the output of all previous steps, even though these previous steps may have been defined at quite different levels.

Ripple-through

The impact of the new step ripples cumulatively through subsequent steps and years.

It is not just the two steps Non-Demographic Growth, and Non-Demographic Growth 1 which show different numbers. Every step following the new step in order of processing is now applied to the higher numbers generated by the new step with its higher assumption of growth. And the rolled over baseline of each year after the first also incorporates the higher numbers of the previous year.