Fig. 4: Analysis of MO signals in the patient population found malaria negative by the reference methods.

a Comparison of MO signals between malaria negative (by LM, PCR and RDT) study participants who reported a recent malaria infection in the two weeks preceding the measurement (n = 219) and those who did not report a recent infection (n = 294). The graph also shows measurements from n = 12 long-term malaria-free volunteers measured during this study, as well as previous data from n = 20 malaria naïve volunteers measured in non-endemic settings using the same protocol. All RMOD measurements were done in triplicate. The distributions were compared using two-sided Mann-Whitney tests resulting in the p-values indicated in the graph. The box-and-whisker plots show the median, the IQR, the 1–99 percentile of the MO signals and the individual data points which fall outside the 1–99 percentile range. b Correlation between LM-based infection positivity rate and MO signals in patients with undetectable malaria in corresponding age groups. LM positivity rate (blue squares) is shown with the error band representing the 95% confidence intervals (CIs) of proportions. MO signals (black circles) are shown as medians per age group with error bars representing 95% CIs of the medians. The dashed black line corresponds to the median background MO signal in the malaria naïve volunteer population. (0.47 mV/V). The x-axis denotes age groups and the number of biologically independent samples (in parentheses, measured in triplicates) per group for MO measurements (n_MO, black) and expert LM (n_LM, blue). c Probability of hemozoin-containing leukocyte presence (P_Hz+) versus MO signal in malaria negative samples. The black line and gray-shaded area are the probability prediction and 95% confidence band of the logistic regression model. The red circles are the binary data for hemozoin-containing leukocyte presence (1) or absence (0).