Abstract
The gathering of high-throughput omics measurements into population-wide databases has created new opportunities for analytical tools in cell biology and medicine. However, deriving meaningful clinical insights from these datasets is non-trivial, and there is an urgent need for computational methods that can interpret such data to solve real medical research questions. Among computational methods for complex data structures, artificial neural networks, which use abstract and non-linear representations of data, have proven transformative. However, the application of such AI models typically requires sample sizes far beyond what is available in most clinical settings. In contrast, there are now large-scale public databases that are, in principle, sufficiently large to support AI-based approaches, but these datasets originate from diverse cell types and experimental conditions. As such, these broad datasets are often of limited direct use for the specific research questions addressed by clinical research groups. To address this, me research develops methods that learn general biological structure from large public datasets and project these insights onto smaller datasets and specific biological questions. In this lecture, I will present my planned research trajectory and discuss how such models can be developed to make large-scale public data useful in applied and personalized biomedical research
