Reactive Determination Calculations

Once a case has been activated and an initial assessment determination has been created, then the items upon which the determination result depends will be established as dependents so that the system may react to changes which may affect it. This is in sharp contrast to traditional case processing which is centered around having to write processing to identify affected cases; instead, the paradigm is much closer to that of a "spreadsheet", which automatically calculates results whenever input values change.

When any data that affects eligibility and entitlement is subsequently changed, an item will be written to a precedent change set, and a request for a deferred process made. When executed, the deferred process will then use the stored dependencies to determine the dependents for the precedent items that were changed and recalculate those dependents. A dependent which relates to a case¹will cause that case to be reassessed.

The Dependency Manager does not really "understand" different types of input data; rather, it just knows that a dependency exists from a determination result value to the various input data values used to calculate it (in the same way as a spreadsheet does not understand the purpose of the data typed into it). From a business perspective, the types of input data change that can affect determination results will typically include:

• changes to case data, e.g. start and end dates;
• changes to personal data, e.g. the correction of a date of birth or the recording of a date of death;
• changes to rate data, e.g. an increase in a benefit payment rate; and
• changes to product configuration, e.g. the introduction of a new period of legislation.

The Dependency Manager's lack of understanding of types of input data is the strength underpinning how reactive determinations work. In a spreadsheet-like way, the Dependency Manager will reassess all cases that are dependent upon the data that has been changed, regardless of whether that change ends up affecting zero, one or very many cases. For some changes to input data, the Engine may reassess cases but find that the overall determination result value has not changed, and in these circumstances no new determination will be stored.