Project Details
Projekt
Assessment of dynamic prognostic scores (D: Dynamic scores)
Applicant
Dr. Erika Graf
Subject Area
Epidemiology and Medical Biometry/Statistics
Term
from 2004 to 2011
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 5470786
In many medical contexts, diagnostic and prognostic scores are used to describe patient risk profiles. Based on the information inherent in a set of patient covariates, they are typically intended to quantify for example the probability that a patient has a specific disease or will die or experience some other event of interest within a specified time range. While there is sound statistical methodology to derive such scores even in situations where the individual risk profile may change in the course of time, the usefulness of such dynamic scores is mostly summarized by inappropriate ad-hoc measures, a fact which may seriously hamper the transfer of new scoring systems into clinical routine. Starting with a suitable formal description of the data structure, we will derive adequate measures for the assessment of prognostic scores with dynamic updates. Special emphasis is put on the interpretability of these measures with respect to the chronological order in the data structure. This research is motivated by a number of clinical problems including data from intensive care units, a study on pre-operative chemotherapy for breast cancer patients, and an epidemiological study on allergy, asthma and respiratory health in childhood. The methodology to be developed will be fully applicable also to diagnostic scores with dynamic updates. However applications within this project will be concentrated to the prognostic setting.
DFG Programme
Research Units
