Article
Biomarker selection in profiles of varying dimensionality
Search Medline for
Authors
-
Hans Kestler
- Research Group Bioinformatics and Systems Biology, Ulm University, Ulm, DE
| Published: | August 27, 2013 |
|---|
Outline
Text
Introduction: In recent years, high-throughput methods have become standard tools in clinical investigations. Specialized machine learning methods allow for the analysis of these high-dimensional patterns for identifying biomarker combinations that are related to specific phenotypes. Such marker combinations can aid in the identification of new disease subtypes and provide decision support for diagnoses.
Methods: The talk presents supervised machine learning techniques that target specific issues in clinical studies. It covers the identification of relevant biomarkers in high-throughput scenarios as well as specialized classifiers. In particular, a classifier that chooses representative subjects that are prototypic for certain subgroups is introduced.
Results: The new methods showed their usefulness in different clinical studies. Theoretical results show that such methods are well-suited for clinical datasets of low cardinality.
Discussion: Depending on the particular study setting, the employed methods must meet different requirements. The talk highlights some particular aspects and proposes suitable approaches. In practice, these may be further adapted to the particular requirements of a study.
References
- 1.
- Guyon I, Gunn S, Nikravesh M, Zadeh LA. Feature Extraction: Foundations and Applications. Springer; 2006.
- 2.
- Lausser L, Müssel C, Kestler HA. Measuring and visualizing the stability of biomarker selection techniques. Computational Statistics. 2013;28(1).
- 3.
- Müssel C, Lausser L, Maucher M, Kestler HA. Multi-objective parameter selection for classifiers. Journal of Statistical Software. 2012; 46(5):1–27.
- 4.
- Meyer LH, Eckhoff SM, Queudeville M, Kraus JM, Giordan M, Stursberg J, Zangrando A, Vendramini E, Möricke A, Zimmermann M, Schrauder A, Lahr G, Holzmann K, Schrappe M, Basso G, Stahnke K, Kestler HA, Te Kronnie G, Debatin KM. Early
relapse in ALL is identified by time to leukemia in NOD/SCID mice and is characterized by a gene signature involving survival pathways. Cancer Cell. 2011;19(2):206-17. DOI: 10.1016/j.ccr.2010.11.014
- 5.
- Kestler HA, Lausser L, Lindner W, Palm G. On the fusion of threshold classifiers for categorization and dimensionality reduction. Computational Statistics. 2011; 26(2):321–340.
