Connecting the Dots: Multimodal Integration in Healthcare through Multilayer Networks

• 
  Tuesday, 4 March 2025
• 
  11:30h - 11:45h
• 
  • 

    Davide Cirillo
    Leading Researcher at Life Sciences Department, Barcelona Supercomputing Center (BSC)

  • 

    Aparajita Haldar
    Senior Researcher in AI and Large Graphs at Fujitsu Research

• 

  Healthcare data is collected in multiple modalities (e.g., histopathology slide images, genomics, clinical health records) that must be integrated to facilitate a cohesive understanding of the patient’s disease status and to offer bespoke treatment therapy options. To this end, we present some outcomes from a research collaboration between Barcelona Supercomputing Center (BSC) and Fujitsu Research of Europe Limited  (FRE).

  Our BioFusion package is aimed towards the integration of a gene association multilayer network (MLN) and AI-based multi modal (MM) graph learning technologies for better performance and explainability of downstream tasks. BioFusion achieves state-of-the-art results on tasks of cancer diagnosis, and uses the per-patient MM outputs together with the global MLN gene relationships to produce visualisations and genetic insights.