Print This Page

Cletus Kwa Kum
-- licentiate thesis --

Bayesian Analysis of Two Malaria Treatments and Probability Modelling of Malaria Parasite Genotypes

Abstract
We analyze data from a two-centre study of malaria treatment of children in Tanzania from two aspects. The efficacy of two drugs is evaluated using Bayesian statistical methods. Simple models on the curative ability as well as the ability to delay reinfection are formulated and the posterior distributions are obtained. The drug in combination with artemisinin-derivative is seen to be a better treatment regimen. These results confirm existing knowledge that Plasmodium falciparum offers less resistance to Artemisinin Combination Therapy. Probability models are formulated to estimate the probability of infections with one or several types of genotyped malaria parasites. We study three genes which are suspected to exists in two forms: susceptible and resistant. One of the susceptible genes is affected much while for another locus, the ratio between susceptible and resistant is not affected. Separate models at baseline, first instance of reinfection and at both time points are considered. The superior models that provide a better fit to data are used to measure the effects of treatment on parasite in the genotypes.

Keywords: Bayesian analysis, Efficacy, Infection probability, Malaria genotyping, Susceptible gene, Resistant gene.