Fig. 6.

Representative communities with population mean malaria incidences ranging from 0.5 to 4.0 episodes per person-year were created by subsampling actual study participants after Poisson distributions. Analyses of actual participant data for individuals assigned to each of these simulated communities showed that serologic analysis of a small number of individuals produced accurate predictions of mean incidence for the community. Predicted community incidences were aggregated from cross-validated estimates for each individual, which were produced using antibody responses to six antigens. More details are available in Materials and Methods.