Institute of Medical Biometry and Statistics (IMBI)

This project is completed and this page is archived. Last change on this page was 2013.

Question

Improved AML Patient Management Based on Risk Prediction Models Incorporating Molecular Biological and Clinical Dependencies

Summary

Although numerous of clinical relevant molecular markers are identified in patients with acute myeloid leukemia (AML), there persists a huge heterogeneity in prognosis within the corresponding molecular subgroups. One way to improve therapy management is to consider risk prediction models, which integrate multiple genome wide data sets with regard to clinical endpoints as therapy response and survival times. Therefore, novel statistical procedures for an integrated analysis of genome wide data sets for the AML problem are adapted. Such procedures should strive to identify correlated quantities from two or multiple data sets, e.g. SNP measurements and expression measurements. Most of these procedures do not allow for simultaneously relating these measurements to clinical endpoints, therefore approaches specific for this setting will be developed within this project. This promises an improved prediction concerning therapy response and prognosis. Subsequently, the results of these novel developed procedures are experimentally verified. Thus, in addition to novel insights into the molecular mechanisms, a basis for improved therapy management is created.

References

• Harald Binder, Axel Benner, Lars Bullinger, and Martin Schumacher, "Tailoring sparse multivariable regression techniques for prognostic single-nucleotide polymorphism signatures", Statistics in Medicine, 2012.
• Stefanie Hieke, Axel Benner, Richard F. Schlenk, Martin Schumacher, Lars Bullinger, and Harald Binder, "Intergation of multiple genome wide data sets in clinical risk prediction models", BMC Bioinformatics, 2013. [Submitted]
• Stefanie Hieke, Axel Benner, Richard F. Schlenk, Martin Schumacher, Lars Bullinger, and Harald Binder, "Componentwise boosting vs. univariate testing: a comparison of strategies for clinical cohort SNP data", Bioinformatics, 2012a. [Manuscript]