30 July 2014

Dr. Sach Mukherjee, Programme Leader at the MRC Biostatistcs Unit & School of Clinical Medicine, University of Cambridge, UK 

Venue: B-IT Lecture Hall, Dahlmannstraße 2, 53113 Bonn 
Time: 16:0 30 hours 

Abstract: Disease subspaces: high-dimensional approaches to explore disease heterogeneity

Human diseases, including cancer, show considerable heterogeneity at the molecular level. Such heterogeneity is central to personalized medicine approaches that seek to exploit molecular data to inform clinical decision making and improve patient outcomes. However, heterogeneity leads to nontrivial challenges in the analysis and interpretation of molecular data, due to the fact that different diseases and disease subgroups may live in quite different data or parameter subspaces. For example, links between molecular variables or associations between molecular variables and clinical outcomes may themselves depend on disease subgroup. This observation has implications for a number of machine learning and statistical approaches, including clustering, imensionality reduction, hypothesis testing and prediction. I will describe our ongoing efforts to develop high-dimensional methods that can be used to explore disease heterogeneity in a robust yet truly multidimensional manner. I will illustrate the methods in the context of the Cancer Genome Atlas "Pan-Cancer" study and point to a number of open questions and challenges in this area.

Upon the invitation of Professor Armin B. Cremers, Dr. Sach Mukherjee, Programme Leader at the MRC Biostatistcs Unit & School of Clinical Medicine, University of Cambridge, UK, delivered a well attended and very insightful talk about "Disease subspaces: high-dimensional approaches to explore disease heterogeneity".