Computational cancer biomedicine

Problem

The immune system is an extremely sensitive warning sensor, scanning for infections and cancer. Its responses often decide whether we recover or fall seriously ill, yet we still lack ways to use this information to guide treatment. In oncology, we still rely on century-old blood tests (blood counts, CRP) that provide only crude insights. New single-cell technologies now allow us to examine immune cells in unprecedented detail. Combined with artificial intelligence, they could reveal how cancers arise, how they can be intercepted early, and how therapies can be tailored to each patient.

Approach

We address this gap by linking clinical trials, experimental systems, and artificial intelligence. Leveraging prior knowledge and large data repositories enables us to build interpretable AI tools that generalize to thousands of patients, enabling detailed views of how the immune system eliminates or fosters tumors. We validate these insights using patient-derived experimental systems. Multimodal data integration frameworks allow us to integrate these insights into existing artificial intelligence workflows for electronic health records, radiology, and pathology data. Link to GitHub  

Impact

We previously established that we can predict clinical outcomes from blood immune responses. We now leverage these insights to build a new generation of immunodiagnostics that guide therapy with precision which we call artificial intelligence immunodiagnostics. These tools could distinguish whether inflammation reflects infection, autoimmunity, or cancer determining if a patient should receive antibiotics, immunosuppressive treatment, or adjusted cancer therapy. Beyond diagnostics, our research highlights new targets to strengthen anti-tumor immunity and to intercept malignant disease early.