Fig. 3

The window of selection (WoS) for artemether–lumefantrine (AR-LF). Data from 5,000 patients; the earliest successful emergence from the liver (which determines the ‘true’ WoS) is shown in a, the earliest observed patent infection (used to calculate the observed, ‘clinical’ WoS) determined using Method 1 is shown in b and using Method 2 is shown in c. The histogram and box plot in each row summarize the same data in different forms. Data plotted in the top row corresponds to the mean LF IC50 and data in subsequent rows shows the results of increasing the LF IC50. Patients were followed up for 365 days with the follow-up time censored when clinical WoS to 63 days, the maximum recommend time of a clinical trial. All new infections occurring after the 63-day follow-up period were either (1) removed from the analysis (b) or (2) re-classified arbitrarily as occurring at day 100 (c).