Cancer – a disease of the genome – is the second leading cause of death globally and is responsible for an estimated 9.6 million deaths in 2018. To make cancer treatments more effective it needs to be personalized from diagnosis to treatment. In today’s clinical practice, however, the information from the genome is either not used or is used inefficiently.

Team Arcas is building an AI-based diagnostic decision support system for cancer. This system makes sense of complex genomic information to support personalized, precise diagnosis and therapy recommendations. At the core, the system analyzes complex genomic information: every cancer biopsy is sequenced not only for mutation detection, but also for large-scale alterations, gene expression, and epigenetic changes. Arcas is using a multi-level deep learning approach to integrate clinical, genomic, and pharmacological data.

With this system, Arcas can predict patient cancer subtypes, survival outcomes, and personalized drug response, more precisely. Arcas has shown promising results for colon, breast, and lung cancer. With more data available, the system can be used for many more cancerous diseases.

The Arcas system also serves pharmaceutical R&D purposes, e.g., by identifying biomarkers to support the stratification of clinical trial participants, or by helping to interpret the molecular differences between responders and nonresponders to a particular pharmaceutical product. Furthermore, Arcas aims to improve cancer diagnostics and clinical therapy decisions. 

Team Arcas consists of international experts in the field of bioinformatics, omics data science, and medicine from the Institute for Medical Systems Biology at Max Delbrück Center for Molecular Medicine in the Helmholtz Association in Berlin. The team is supported by an advisor in life sciences and new venture development.