Logo UHasselt





Logo UHasselt Universiteit Hasselt - Knowledge in action

Centre for Statistics (CENSTAT) : project R-5520

Title : Development of new semi-parametric mixture models for interval censored data: mathematical properties, performance for finite samples, comparison of different approaches, with applications in antimicrobial resistance (R-5520)
Abstract: Antimicrobial resistance has become one of the main public health burdens of the last decades and it is therefore extremely important to study and keep track of the emerging of resistant isolates. This requires the analysis of the minimum inhibition concentration (MIC), defined as the smallest concentration of an antimicrobial that will inhibit the growth of a microorganism. Nevertheless, mathematical developments in the field of antimicrobial resistance (AMR) are very scarce as most attention is paid to the collection of minimum inhibition concentration data rather than to the analysis. In a typical analysis, MIC values are dichotomized using Epidemiological Cut-Off values (ECOFFs) to determine the prevalence of resistant isolates. However, analysis of the MIC data on the binary scale suffer from several shortcomings, one of which being that a trend above the cut-off value cannot be detected. Therefore, it is recommendable to study the MIC distribution on the continuous scale. Mixture models have experienced increased interest over the last few decades and have become well-recognized and popular models. They offer a natural framework for modeling unobserved population heterogeneity and are as such ideally suited for the modeling of AMR data. In this project, my interest is in developing and studying new mixture models applied in the context of AMR. The overall objective is to develop and study the mathematical and finite sample properties of mixture models for AMR data, at the continuous scale, taking additional complexities into account as well as modeling the effect of covariates and time trends. Specific objectives are i) the development and study of the properties of an estimator of a new hierarchical mixture model, ii) the study of its extension to censored data, iii) the study of its extension to model different time trends within the different components of the hierarchical mixture. The study of i)-iii) includes the study of mathematical properties, the study of finite sample properties through simulations, the implementation and development of software, and the application to real life AMR data. The resulting estimator will provide an excellent tool for monitoring the emerging of resistance.
Period of project : 1/10/2014to30/09/2016

Other projects of Centre for Statistics (CENSTAT) :

Details of Centre for Statistics (CENSTAT) : back