Fig 7. Results from generalized additive logistic regression for the geographic, demographic and environmental predictors of scrub typhus infection.

A.) mean NDFI (normalized difference flooding index) for patient home village across study period; B.) mean EVI (enhanced vegetation index) for patient home village across study period; C.) variance in NDFI for patient home village across study period; D.) variance in EVI for patient home village across study period; E.) NDFI around patient home village in 16 days leading up to hospital admission; F.) EVI around patient home village in 16 days leading up to hospital admission; G.) year of hospital admission; H.) day of year of hospital admission; I.) population of patient home village; J.) distance from patient home village to nearest major road; K.) elevation of patient home village; L.) geographic coordinates of patient home village. Plots above the blue line (at the y-intercept) indicate a positive association for a given covariate, for the given x-axis value. For the geographic coordinates (panel L), lighter colored areas on the map indicate places with higher baseline risk of being diagnosed with scrub typhus. P-values are indicated in S5 Table.