Fig. 7.

(A) Geographic heterogeneity in Pf exposure is captured by serologic predictions of incidence. (Left) Average monthly counts of female Anopheline mosquitos, (Center) observed malaria incidence over 1 y, and (Right) cross-validated predictions of incidence using antibody responses to six antigens taken at a single time point are plotted for each study household in Kanungu, Uganda, with colors indicating the tertile for each household. Some households contain more than one included study participant; in these cases, the household mean is plotted for observed and predicted incidence. Small black dots represent households that were not sampled. (B) Scatterplots of (Left) household elevation (a proxy for Pf exposure) vs. mean mosquito counts and (Center) observed or (Right) predicted malaria incidence in the last year. All three metrics are significantly associated with elevation (P < 0.001 for Spearman’s correlation). Individuals residing in households at low elevations but having no episodes of clinical malaria in the last year are likely highly exposed and immune; serologic predictions of incidence suggest that these individuals were, in fact, exposed to Pf.