|
Abstract:
|
Controlled human malaria challenge trials enroll malaria-naïve individuals who are randomized to a placebo or an intervention (drug or vaccine), exposed to or "challenged" with malaria via infected mosquitos or direct inoculation of parasites, and followed for incident malaria infection. Thick blood smear is the current gold standard method of infection diagnosis, but several biomarkers are being investigated, which may improve the diagnosis of infection. Data to evaluate these biomarkers derive from multiple challenge trials of various interventions, where longitudinal biomarker measurements and thick blood smear test results are available post-challenge on each participant. Data are right-censored at the time of thick blood smear positivity at which time participants are treated for malaria infection. We describe a novel approach to quantifying the accuracy of a biomarker for classifying subjects with respect to thick blood smear positivity a fixed number of days later. We describe an approach to estimation and inference, and apply the approach to evaluate the performance of a reverse transcription PCR (RT-PCR) biomarker.
|
