CV

DEGREES

2010 PhD (Health Science) University of Copenhagen, Title: Classification criteria of syndromes by latent variable models: HIV-associated lipodystrophy syndrome

2001 Cand.Scient. (M.Sc) in Statistics, Department of Mathematical Statistics, University of Copenhagen. Title of thesis: An Application of State Space Models

EMPLOYMENT

Head of Copenhagen Phase IV Unit and Section for Biostatistics and Pharmacoepidemiology

Associated Professor Section of Biostatistics, Department of Public Health, University of Copenhagen

Forskningsområde

I am currently working with problems related to the definition of medical syndromes (in particular lipodystrophy) and the use of multiple biomarkers as imperfectly measured indicators for a disease. In both situations, latent variable models - such as Structural Equation Models and Latent Class Regression models - may come in as useful tools for the analyses.

Latent Class Regression (LCR) is widely used within social and psychological research for analyzing variables thought not to be directly measurable, but only possible to infer from surrogate measures. While full LCR modeling is readily implemented using maximum likelihood estimation, it is often preferred not conduct a full LCR for each new analysis, but rather derive scores, which can subsequently be use as outcomes in a variety of analyses. I am right now studying the three-step analytic that follows from such a preference: (i) a latent class analysis is fit without covariates; (ii) the resulting model is used to derive estimates of latent class membership; and (iii) the estimated memberships are analyzed in a regression analysis on the covariates of interest.