Methods to control for confounding variables in screening for associations in the WHO drug safety database
The heterogeneity of the WHO database may lead to problems with confounding (when ancillary variables distort the apparent relationship under study). The aim of this Master's thesis project will be to study the impact of confounding variables in IC analysis on the WHO database and to propose a strategy for how to handle this issue in practice - i.e. when to carry out stratified analyses and how to select the variables to stratify by.
The successful applicant should be a last year student of engineering (civilingenjörsprogram) or mathematical statistics. She or he should have a great interest in mathematical statistics and methods for knowledge discovery in databases, especially as applied to real world problems. Some programming skills are required as all proposed methods are to be implemented and evaluated in practice (we use Matlab for research projects). Both research and development is carried out in a Linux environment, and the official language for the project is English.
For more information or to apply (include CV), please contact:
Data Mining Development
Uppsala Monitoring Centre
Stora torget 3
753 20 UPPSALA
Informationen om uppsatsförslag är hämtad från Nationella Exjobb-poolen.