Our current research projects address various aspects of statistical modeling of medical data.

STRATOS – in this international initiative methods for the analysis of observational data are investigated. Members from our team: Georg Heinze, Daniela Dunkler, Christine Wallisch, Michael Kammer

SAMBA –in this bilateral research project jointly funded by FWF and DFG methods for multivariable modeling of cardiological data are investigated. Principal investigator: Daniela Dunkler

COVID-PRECISE - with a 'living systematic review' models for prediction and diagnosis of COVID-19 are collected and evaluated by an international consortium of experts in prognosis research. Members from our team: Georg Heinze, Christine Wallisch, Maria Haller, Michael Kammer

CaReSyAn - in this EU-funded Innovative Training Network, statistical models for the evaluation of high-dimensional proteomics markers from urine or plasma as predictors of nephrological and cardiological endpoints are investigated. Principal investigator: Georg Heinze

SurvivalBenefit – in this project funded by the Austrian National Bank we quantify the benefit of kidney transplantation using cutting-edge methodology for causal inference. Principal investigator: Maria Haller

PREMA - this bilateral research project jointly funded by the FWF and ARRS investigates models and validation of prediction models for rare events. Examples from this project can be found in the description of software packages for the Firth-correction. Principal investigator: Georg Heinze

Variable selection and shrinkage - the selection of variables for multivariable models and methods to correct for the overfitting and variance inflation implied by this practice have been addressed in review articles (Heinze_et_al- BiomJ; Heinze_et_Dunkler-Transpl Int) and in the ABE and shrink projects.

Survival analysis - in many medical questions the analysis of survival of patients is of main interest. These questions also stimulate our own methodological research. Examples can be found in the description of the software packages kmdiff, SurvCorr, PSHREG, WCM, concreg, compass, XLINRANK, sures, rcs.

Explained variation measures the relative increase in predictive accuracy by adding prognostic factors to a model. Examples of projects on this topic can be found in the description of the  NecSuff, pev_frailty, RELIMP, surev, R2POI, EVLOGIST, Kent&O'Quigley, MULTIMP software packages.

High dimensional data - tests for high dimensional data were covered in the lim.lrt and t.opt projects.