InterventionConditions

Conditional intervention deployment

Here is an example triggered intervention:

  <interventions name="triggered MSAT">
    <human>
      <component id="MSAT">
        <decisionTree>
          <diagnostic diagnostic="RDT">
            <positive>
              <random>
                <outcome p="0.9">
                  <treatSimple durationLiver="0" durationBlood="20d"/>
                </outcome>
                <outcome p="0.1">
                  <treatFailure/>
                </outcome>
              </random>
            </positive>
            <negative>
              <noTreatment/>
            </negative>
          </diagnostic>
        </decisionTree>
      </component>
      <deployment>
        <component id="MSAT"/>
        <condition measure="nTreatments1" minValue="20" initialState="true"/>
        <timed>
          <deploy coverage="0.8" time="19t" repeatStep="1y" repeatEnd="20y"/>
          <deploy coverage="0.8" time="37t" repeatStep="1y" repeatEnd="20y"/>
        </timed>
      </deployment>
    </human>
  </interventions>

This rules out a mass-screen-and-treat campaign twice a year (steps 19 and 37) every year for 19 years, conditional on there being at least 20 first-line treatments given during the last completed survey. See full example.

The line

        <condition measure="nTreatments1" minValue="20" initialState="true"/>

specifies that deployment is conditional on measure nTreatments1 having value at least 20 (minValue=...). It is also possible to specify maxValue=... (with or without the min value). minValue and maxValue define closed intervals for intervention triggering (if the measure is greater than or equal to minValue and less than or equal to maxValue).

Conditions are only updated when a survey completes. The initialState specifies the state of the condition before the first survey in the simulation completes. When initialState="true", deployment occurs if the (start) time is before the first survey. To allow conditions to be updated without reporting a full survey, unreported surveys can be used.