Feature selection in high dimensional data for risk prognosis in oncology (C01)

• Subject Area: Bioinformatics and Theoretical Biology ; Hematology, Oncology
• Term: from 2011 to 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 124020371

Reliable interpretation of very high-dimensional data with limited sample size is and remains a yet unsolved challenge for machine learning and data analysis. Robustness of feature selection and prediction is an important consideration. The long-term goal is to generate reliable prediction models for precise definition of patient risk in oncology using neuroblastoma, a common solid tumor of childhood, as a model. We will now proceed to develop probabilistic graphical models on the basis of next generation sequencing data and other high-throughput data not only for better interpretability of feature selection but also to analyze tumor development over different time points.

• DFG Programme: Collaborative Research Centres
• Subproject of: SFB 876: Providing Information by Resource-Constrained Data Analysis
• Applicant Institution: Technische Universität Dortmund
• Project Heads: Professor Dr. Johannes Köster, since 4/2021; Dr. Sangkyun Lee, from 1/2015 until 2/2017; Professorin Dr. Katharina Morik, until 12/2014; Professor Dr. Sven Rahmann, from 1/2015 until 3/2021; Professor Dr. Alexander Schramm