Systems biology is a branch of the life sciences, which tries to understand biological organisms as a whole. Systems biology is rich, wide and diverse area of research, where the study of biological networks, such as genetic and metabolic networks, is of central interest. The study of biological networks includes the study of the network components and their interactions, which result in a dynamic behavior. The task of reconstructing biological networks, including their components, structure/interactions and dynamic behavior, from time course data is at the heart of systems biology.

In this talk, we present the use of process models for modeling the dynamics of biological networks. Process models are formalism for modeling dynamic systems that integrates qualitative and quantitative aspects of the observed system dynamics. The qualitative part of the model, that takes form of entities and processes, explains the dynamic behavior, while the quantitative part, that takes form of differential equations, allows for simulation of the model. Thus, process models provide an appropriate formalism for integrating the graphical presentation of biological networks, as qualitative structure of components (entities) and interactions (processes), with the dynamic network behavior, that can be inferred by the quantitative part of the model. We demonstrate the use of inductive process modeling on illustrative tasks of crafting the complete dynamics of biological networks, such as rediscovering known networks and proposing models for unknown networks.

Statistical methods for meta-analysis can be used to combine and compare results from independent studies designed to make inference on the same scientific issue. In environmental epidemiology research, meta-analysis is commonly employed to estimate overall measures of effect or to identify inconsistency among results arising from different locations within multi-centre studies. Examples concern the study of the short term effect of urban air pollution on mortality and morbidity and the effect of warm and cold temperatures on health.
In this context, because of regional, demographic and environmental differences among locations, the assumption of homogeneity of effect measures across trials is usually inappropriate. In the presence of heterogeneity, random effects meta-analysis can be used, but the interpretation of the combined effect estimate is not straightforward and its internal and external validity falls.
This is a relevant problem when meta-analytic results are used for health impact assessment purposes. Use of shrunken estimates, which are a compromise between overall and location-specific estimates, is recommended in these situations.

Relative survival methodology has a long history, which dates back to the 1950's. Its primary objective was to judge in a standardised manner the therapeutic progress, freed from the subjective determination of the cause of death of the patients. The pioneers of the methodology had clearly in mind the estimation of the net survival. For many years the methodology was based on the computation of the life table expected survival. This methodology has been used up to now by the cancer registries for the evaluation of the survival of the incident cancer patients.

In parallel important advances was made in the study of the statistics of stochastic processes and in modelling survival with multivariate regression. It was natural to try to model relative survival in the same way and much progress has been made in that respect.

This seminar will discuss the basic concept of net and relative survival and will retrace the evolution of the methodology. It will be illustrated with several examples showing the flexibility of the new methodology.

In practice the new tools are not widely used by cancer registries, despite their advantages over the classical estimates based on the computation of expected survival. It is likely that the lack of easy-to-use procedure within the more commonly used statistical software is responsible for this anomaly. The example of the "relsurv" R-package developed by Pohar et al should be followed by the development of similar tool for commercially available statistical software.

Multi-state models are typically used when subjects, e.g. patients, are followed over time and a number of events may be observed for each subject. In such studies, the occurrence of an event is modelled as a transition between two states in the model. Bone marrow transplantation is one area where multi-state models have been used extensively. Here, the events of interest both include ultimate events like death and relapse and transient events like graft versus host disease. We will give a review of such methods with emphasis on regression models for both transition intensities and transition- and state occupation probabilities. Both semi-parametric models, like the Cox regression model, and parametric models based on piecewise constant intensities will be discussed. Illustrative examples will be taken from studies in bone marrow transplantation.

R is a versatile statistical programming and analysis platform. LaTeX is for decades one of the most efficient typsetting environments for technical and mathematical texts. Sweave is a glue that binds the two into even more effivcient bundle.

Sweave was developed by F. Leisch and is a part of standard R. Sweave follows te idea of 'noweb' which enables the combination of comments and program code in one document. This document, which closely resembles LaTeX file combines elements of final report and program code that produces analysis results. Results in a form of tables and graphs are constituent part of the report and are automatically included in the final text. In this sense Sweave is a system compliant with the paradigm of 'reproducible research': analysis and comments are combined in one file that documents the analysis itself.

In the lecture we will show the basic features and use of Sweave. We will give the practical advices how to start using this efficient system for reports and documentation of statistical analyses using R.

Econometrics is a relatively young discipline. We will give a short introduction to its origin and field of research. Its classical tool is multiple regression analysis, which is also the base for development of many new methods and approaches in econometrics.

Econometrics is often concerned with the study of many economic principles (also in health economics) based on so called cross-sectional data. We will in short explain the basic problems of heteroscedasticity in classical OLS models. A few approaches for testing for heterescedasticity will be presented and procedures given to get more efficient estimator than OLS.

Statistical methods and approaches in business (i. e. manufacturing and service) environment will be very concisely presented. The main stress will be laid upon Statistical process control, a field traditionally of central importance and further emphasized by the advent of ISO 9000 series of business standards. Other fields will be mentioned, too, notably Design of experiments with selected examples of application. Some comments will be given on the role and importance of statistics in business environment in Slovenia and on its standardization and terminology.

The goal of many gene-expression microarray profiling clinical studies is to develop a multivariate classifier to predict patient disease outcome from a gene-expression profile measured on some biological specimen from the patient. This lecture will introduce the most commonly used class prediction methods, show how to develop a classifier and how to assess its predictive accuracy. Examples from the literature will be used to point out some common errors.

Six graduate students of statistics (Darja Sušnik, Vanja Govednik Erculj, Sandra Turk, Rudi Seljak, Aleš Žiberna, and Tomaž Štupnik) talked about their work in connection with statistics, critically assessed the good and not so good sides of the graduate programme, and suggested possible improvements.

OPOZORILO : Pregledujete staro stran IBMI

Vsebine na strani so zastarele in se ne posodabljajo veÄ. Stara stran zajema doloÄene Älanke in vsebine, ki pa morajo biti ĹĄe vedno dostopne.

Za nove, posodobljene vsebine se obrnite na http://ibmi.mf.uni-lj.si/

Events in 2008

OPOZORILO : Pregledujete staro stran IBMI