Explore effect of interventions

Interventions are steps without IF condition (start steps) that only appear in other IFNOT conditions, i.e., that can only prevent steps but not cause them. Interventions are not considered when creating SCC models using create_scc(). intervene() evaluates their impact in two directions: 1) which sufficient causes can be prevented by certain (sets of) interventions and 2) which set of interventions is at least needed to prevent the outcome in an individual with a given set of component causes.

Usage

intervene(scc, causes = NULL, intervention = NULL, output = c("nice", "table"))

Arguments

scc: An object of class epicmodel_scc.
causes: A character vector containing step IDs of component causes. If "all", investigates all sufficient causes, i.e., all minimally sufficient sets of component causes. If NULL (default), prints a list of all available component causes in the console. If a set of step IDs is specified, only the specified set is investigated.
intervention: A character vector containing step IDs of interventions. If "all", investigates all possible combinations of available interventions. If NULL (default), prints a list of all available interventions in the console. If a set of step IDs is specified, investigates all possible combinations of the specified interventions.
output: Either "nice" (default) or "table". If "nice", prints a nicely formatted summary in the console. If "table", returns a list of several elements described in detail in section "Value" below.

Value

Output

If output = "nice" (default), prints a nicely formatted output in the console. If output = "table", returns a list with the following elements:

cause_set: A list of character vectors with one element for every investigated set of component causes. The character vectors contain the step IDs of the component causes that are part of the corresponding set. Sets are named in a format similar to cc1, cc2, etc.
intv: A list of character vectors with one element for every investigated set of interventions. The character vectors contain the step IDs of the interventions that are part of the corresponding set. Sets are named as intv1, intv2, etc.
status: A data.frame with one row per set of component causes and one column per set of intervention. In addition, contains one column representing no interventions (intv0). Each cell contains the sufficiency status of the corresponding set of component causes when the corresponding set of interventions is applied. Possible values are "always", "depends", and "never". See below for an interpretation.
minimal: A data.frame with one row per set of component causes and one column per set of intervention. Each cell is either TRUE or FALSE indicating if the set of interventions is minimal. For non-minimal sets of interventions, a smaller set which is contained within the corresponding set exists and has the same preventive power. Minimality is defined separately for every set of component causes. If both the larger non-minimal and the smaller minimal set sometimes prevent the outcome (status "depends" in status (see above)), the non-minimal set might actually prevent more sufficient orders of occurrence than the minimal set. In this case, please inspect and compare element order (see next), for all minimal and non-minimal sets of interventions with status "depends".
order: A 2-level list, i.e., a list with one element per intervention set, for which each element is another list with one element per evaluated set of component causes. Each intervention/component causes combination contains a data.frame, similar to the data.frames in the sc_order element of epicmodel_scc objects, if the corresponding status is "depends", or is NA otherwise (for "always" or "never"). The data.frames contain two columns, which are called "order" and "suff" (short for "sufficient"), and one row for every order of occurrence. The order of occurrence is summarized in "order" (as character), while "suff" is either TRUE or FALSE indicating if the corresponding order of occurrence is sufficient, i.e., leads to the outcome, or not. Please note that the prevented orders of occurrence have suff == FALSE.

How to interpret `status`

If the sufficiency status for a certain intervention in column intv0 is always, the three sufficiency status options for a certain intervention have the following interpretations:

always: The corresponding set of inteventions never prevents the outcome, because after applying the intervention, the corresponding set of component causes is still always sufficient.
depends: The corresponding set of interventions sometimes prevents the outcome, because after applying the intervention, sufficiency for the corresponding set of component causes depends on the order of occurrence.
never: The corresponding set of interventions always prevents the outcome, because after applying the intervention, the corresponding set of component causes is never sufficient.

If the sufficiency status for a certain intervention in column intv0 is depends, the sufficiency status options for a certain intervention have the following interpretations:

depends: The corresponding set of interventions sometimes or never prevents the outcome, because after applying the intervention, sufficiency for the corresponding set of component causes depends on the order of occurrence. Further inspection and comparison of sufficient orders of occurrence is necessary to determine if the intervention actually prevents anything.
never: The corresponding set of interventions always prevents the outcome, because after applying the intervention, the corresponding set of component causes is never sufficient.

If the sufficiency status for a certain intervention in column intv0 is never, no intervention is necessary, because the corresponding set of component causes is never sufficient.

Details

The following algorithm is used to evaluate the effect of interventions:

Derive the list of intervention sets to evaluate
Derive the list of sets of component causes to evaluate
Evaluate sufficiency without intervention for every set of component causes
Evaluate sufficiency for every combination of intervention set and set of componen causes: First, check which steps are prevented by the corresponding set of interventions, i.e., for which steps the IFNOT condition is fulfilled by the intervention set. These steps are removed from the list of available steps. Second, evaluate sufficiency based on the remaining steps similar to create_scc() (Check sufficiency & Check IFNOT conditions).
Evaluate, which intervention sets are minimal, i.e., at least necessary to prevent the outcome

Examples

# Derive SCC model
scc_model <- scc_rain

# Inspect the effect of interventions
intervene(scc_model, causes = "all", intervention = "all")
#> 
#> ── Intervention ────────────────────────────────────────────────────────────────
#> 
#> ── Cause Set 1 ──
#> 
#> • rain
#> • get groceries
#> 
#> Status without intervention
#> ✔ Always sufficient
#> 
#> Status with intervention
#> ✔ Complete prevention by the following minimal intervention sets
#> 
#> ── Intervention Set 1 
#> • take umbrella
#> 
#> ── Cause Set 2 ──
#> 
#> • no vacation
#> • weekday
#> • rain
#> 
#> Status without intervention
#> ✔ Always sufficient
#> 
#> Status with intervention
#> ✔ Complete prevention by the following minimal intervention sets
#> 
#> ── Intervention Set 1 
#> • work from home
#> 
#> ── Intervention Set 2 
#> • take umbrella
#> 
#> ── Intervention Set 3 
#> • take vacation
intv <- intervene(scc_model, causes = "all", intervention = "all", output = "table")