Doktorand/utbildningsplats

Sahlgrenska akademin utlyser doktorandplatser med placering vid institutionen för biomedicin, institutionen för medicin och institutionen för neurovetenskap och fysiologi.

Doktorandplatserna avser två års utbildningsbidrag och två års doktorandanställning, alternativt fyra års doktorandanställning, och beräknas leda fram till doktorsexamen.

För mer information och ansökningsblankett se:
http://www.sahlgrenska.gu.se/doktorandportalen/Doktor_Fran_start_till_mal/doktorandplatser


Regression analysis for log normal data in biomedical research.Development and evaluation of biostatistical methodology

Application deadline: 2010-09-21

Description of the research project
Data in biomedical research can often be described well by a log normal distribution, i.e.a skewed distribution with many low values and fewer high ones.The log normal distribution is biologically reasonable, e.g.for biomarkers, risk- and health factors or exposure data.It is often of interest to investigate which variables are associated with a response variable (e.g.a biomarker).In many situations it is reasonable to assume a linear relationship between the response variable and the explanatory variables: the expected value of the response increases linearly with the explanatory variables.In a situation with a linear relationship and a log normal distribution, the log transformation will distort the linear relationship.We need to develop new methods for estimating a linear relationship in situations where the response variable follows a log normal distribution.Since many variables in medicine and biology can be described with a log normal distribution, appropriate methods in this area is of great importance.The Sahlgrenska Academy has recently decided to focus more on epidemiology and biostatistics, and to devote special resources for a Center for epidemiology, bio statistics and clinical studies.An important part of this is to improve and deepen the competence in biostatistics, which in turn requires research and method development in this specific area.
The main goal of this project is to investigate a newly proposed method for estimating a regression and test its parameters, using the methodology of maximum likelihood.The purpose is to develop and evaluate the new methodology.The project aims at developing computer programs for estimation, testing and computing confidence intervals, as well as evaluating the properties of the method (bias and variance) and suggesting methods for dealing with small samples and comparing this new method to traditional regression analysis.

Desired background
It is desirable that the candidate has profound knowledge of statistics.