(2) Computational epidemiology of neurodegeneration
[PIs: Amunts, Breteler, Griebel, Klein, Mukherjee, Reuter, Schultze, Schweitzer]
The field of epidemiology addresses questions of etiology and disease risk at the scale of human populations. Dramatic technological advances spanning many modalities, from imaging to genomics, have meant that large and especially high-dimensional data are emerging as key drivers for future advances in human subject research. The neurodegenerative diseases are associated with enormous human and economic costs and represent a profound societal challenge. Furthermore, the nature of these diseases, characterized by complex multi-factorial etiology, biological changes taking place over many years, subtle subclinical features and heterogeneous clinical presentation, poses challenges for research methodology and many of these challenges are shared with a range of other complex, mid- to late-life diseases. Motivated by these scientific and societal challenges we propose an ambitious research program located at the interface between epidemiology and the computational data sciences, bringing to bear a breadth and depth of expertise that goes much beyond what has been attempted to date. Our efforts will span, among other areas, high-dimensional statistics and machine learning, image analysis and visual computing, integrative genomics and HPC, and involve world-leading experts in these fields. These areas will be developed in the context of, and integrated with, core questions of etiology and risk prediction in the epidemiology of neurodegenerative disease. Our efforts will centre around the Rhineland Study at the DZNE in Bonn (led by Breteler), a path-breaking study in which up to 30.000 individuals are being studied in a longitudinal manner over 30 years, with deep phenotyping via magnetic resonance imaging (MRI), genomics, epigenetics, metabolomics, clinical/lab tests and more. This represents one of the worldwide deepest phenotyped cohorts currently under study. While computational methods are of course routinely used in parts of epidemiology and biostatistics, what we propose goes much beyond standard interdisciplinary projects in these fields, towards a deeper synthesis with the broader computational sciences. A central emphasis will be the need to move towards truly high-dimensional analysis that takes advantage of emerging data streams to account for many factors and their interplay in a unified fashion. Doing so requires breaking new ground in several areas and at more than one interface, as detailed in the proposal. During the project period we expect to provide impact in the specific areas that we work on, but also to lay the groundwork – conceptual and practical – for truly high-dimensional, next-generation human subject research. This section of RU-B will rely on close collaboration with investigators from RU-C and various topical units. Moreover, it also raises broader epistemological questions that we will pursue within the framework of the CST.
