Fig. 4

Comprehensive plots of the model output to determine efficacy. Each column presents successive transmission cycles 1, 2, 3 and 4 using the 2 mosquito bites per host group as an example. The generated posterior distributions from the model are the most probable values for parasite density for each of 2000 iterations. a–d Histograms of the estimated effect size for atovaquone 32% (ATV-32%) against prevalence (TBA) in the mice generated from the model posterior distributions. e–h The posterior predictive outputs for control (red) and treatment (black) data on the log scale + 1 for each transmission cycle (i). Where points fall below the blue line there has been a reduction in the parasitaemia from the initial (i) to the subsequent (i + 1) cohort of hosts. The controls become both more numerous and dispersed in comparison to the treatment estimates as transmission progresses from transmission cycle 1–4; the drug has a positive impact both by reducing cases and by reducing the amount of infection through successive transmission cycles. i–l Histograms of the estimated effect size for ATV-32% against parasite density (TRA) in the mice generated from the model posterior distributions. For a–d and i–l, the red solid vertical line marks the mean estimate and the dashed blue lines indicate the median estimate