Robust Scaling
EIR - PR
Background
The data set:
A long time ago, a study collected paired estimates of the EIR and the PR [1]; some of this was a thesis [FanelloC2000RelationshipEntomological?].
Then Simon worked on it [2]
We took a mechanistic approach [SmithDL2005_EIRvPR?,SmithDL2007_R0?]
Then we redid it once [3]
Then we redid it again [4]
- not me but also useful [5]
Variability
Among the local factors affecting the relationship between the average annual EIR and the average annual PR are:
Drug Taking
Seasonality – the relationship between the average annual EIR and the average annual PR is lower, but the orbits all like to the left of the average curve.
Spatial Heterogeneity – mosquito populations are highly spatially heterogeneous; this gets averaged out by human mobility patterns
Heterogeneous Biting – some people get more bites on average than others; we handle this with biting weights
Environmental Heterogeneity – exposure rates are highly variabile even with a population stratum; variance around the mean gives negative binomial exposure rates
Travel Malaria
Measurement Errors
Differences in the PfPR by study design: age, sex, location and time of year
Differences in the PfPR diagnostic method
Differences in the mosquito sampling strategy