Scientific rationale:
Current research highlights the need to explore the long-term consequences of infectious diseases like COVID-19, which in many cases are chronic symptoms such as fatigue, cognitive impairment, and other long-lasting health problems. These post-acute infection syndromes (PAIS) are scientifically recognized, but the pathomechanisms are not understood. To develop effective treatments and improve the quality of life of those affected, it is crucial to determine which subtypes of PAIS are caused by which pathogens and to identify involved pathways and biomarkers corresponding to these subtypes.
Objectives:
We aim to comprehensively investigate and understand the underlying molecular mechanisms of PAIS by applying state-of-the-art machine learning and statistical models on clinical, imaging and omics data. We focus on the development of advanced methods and software tools with multimodal integration on large-scale high-dimensional data as necessitated by the complexity of PAIS. Specialized machine learning and omics analysis techniques, including deep learning-based feature and base models, semi-supervised and positive-unlabeled learning, GWAS and QTL mapping, aim to identify unrecorded cases, subtypes, and corresponding genetic markers of PAIS. The obtained insights help to improve diagnostic accuracy, develop new therapeutic approaches, and promote a deeper understanding of disease mechanisms. Specific biomarkers help to better characterize PAIS and determine patient-specific subtypes in the future, significantly improving diagnosis and treatment of PAIS in the medium to long term.
Research questions:
How can the identification of unrecorded PAIS cases be improved through machine and deep learning models?
How can subtypes involving various pathogens and pathomechanisms be identified using the different methods?
What are the characteristics of the identified PAIS subtypes, their corresponding biomarkers and potential therapeutic targets?
