Bayesian analysis of an epidemiologic model of Plasmodium falciparum malaria infection in Ndiop, Senegal

Abstract

Plasmodium falciparum has a complex transmission cycle. Public health planning and research would benefit from the ability of a calibrated model to predict the epidemiologic characteristics of populations living in areas of malaria endemicity. This paper describes the application of Bayesian calibration to a malaria transmission model using longitudinal data gathered from 176 subjects in Ndiop, Senegal, from July 1, 1993, to July 31, 1994. The model was able to adequately predict P. falciparum parasitemia prevalence in the study population. Further insight into the dynamics of malaria in Ndiop was provided. During the dry season, the estimated fraction of nonimmune subjects goes down to 20% and then increases up to 80%. The model-predicted time-weighted average incidences contributed by nonimmune and immune individuals are 0.52 cases per day and 0.47 cases per day, respectively. The median times needed to acquire infection (conversion delay) for nonimmune and immune individuals are estimated at 39 days and 285 days, respectively.