Our group focuses on translational research for breast cancer (BC), which is the most common malignancy and the leading cause of cancer death in women worldwide. We have a widely interdisciplinary spectrum of expertise, consisting of biologists, medical doctors, bioinformaticians and artificial intelligence scientists. We develop computational pathology tools implemented with conventional machine learning and modern deep learning methods relying on large in-house databases of digitized BC tissue slides, genomic and clinical data to extract relevant biomarkers for advancing BC prognostication and therapy prediction. This proposal aims to confront open questions on a) understanding the tumor microenvironment and its heterogeneity with respect to different subtypes of BC, b) investigating the prognostic value of novel AI models with respect to adjuvant/neoadjuvant therapies and c) investigating the prognostic benefit of integrating multimodal data (H&E slides, immunofluorescence slides, tissue morphology, genomics, radiomics and clinical variables). The assigned resources of SNIC Alvis will be utilized to develop and validate these GPU-accelerated AI models in a data-driven way on large international and multi-institutional private and public cohorts.