This project is completed and this page is archived. Last change on this page was 2010.
Dynamic scores
Assessment of dynamic prognostic scores
Duration: 2004-2010
Summary
This project is concerned with providing a general framework for the correct assessment of dynamic prognostic scores. A prognostic score in general is a summary of patient characteristics that bears prognostic information regarding an endpoint of interest, often a survival time. It can be model-based or an 'expert guess'. The dynamic aspect is introduced when patient data is not only collected at a defined starting point of an observation period, but also at later time points when the observation (e.g. the clinical trial) is already ongoing. The repeated calculation of the prognostic score then leads to a dynamic, or updated, score.
The correct chronological treatment of prognosis time and prognosis horizon complicates the assessment and has led to ad--hoc and sometimes even incorrect approaches that unfortunately are used far too often in the applied clinical literature. In the second funding period, the project aims to pursue a unified assessment approach and compare it with the assessment measures developed so far by the project and with approaches proposed in the literature. Comparisons are to be made on a population level as well as with real data.
This research is motivated by a number of clinical problems of our cooperation partners where an urgent need to assess prognostic models using updated patient covariate information has been expressed. This includes data from intensive care units, data on elderly patients from internal medicine wards and breast cancer patients.
Publications
- Schoop R, Schumacher M, Graf E. Measures of prediction error for survival data with longitudinal covariates. Biom J 2011; 53(2):275–293.
- Schoop R, Beyersmann J, Schumacher M, Binder H. Quantifying the predictive accuracy of time-to-event models in the presence of competing risks. Biom J 2011; 53(1):88–112.
- Binder H, Graf E. Encyclopedia of Medical Decision Making, chap. Brier scores. SAGE Publications, 2009.
- Schoop R, Graf E, Schumacher M. Quantifying the predictive performance of prognostic models for censored survival data with time-dependent covariates. Biometrics 2008; 64:603–610.
- Schoop R, Predictive accuracy of failure time models with longitudinal covariates, Dissertation 2008.
- Graf E. Survival and Event History Analysis, chap. Explained variation measures in survival analysis. John Wiley & Sons: New York, 2006.
- Graf E. Encyclopedia of Biostatistics, chap. Explained variation measures in survival analysis. Third edn., John Wiley & Sons, Ltd: Chichester, 2005; 1856–1858.
Principal investigator
Dr. Erika Graf (IMBI)