Network Bio is building a multi-institution research network by connecting top US academic biobanks—Mass General Brigham, the University of Pennsylvania, Duke, and the University of Colorado Anschutz—to create disease-focused, scalable datasets that no single center could assemble alone.

The platform combines tissue, blood, and molecular data with longitudinal outcomes using a bio-native AI architecture, enabling transfer of biological principles across diseases, tissues, and data modalities.

The CEO paints a vision of reading disease barcodes in tissue at scale as the future of medicine, underscoring the ambition behind Network Bio’s approach.

The strategy emphasizes uncovering biological principles transferable across conditions so models grow more capable as more questions are asked, capturing connections across tissues and systems.

The goal is an AI infrastructure that learns universal biological principles rather than building a separate model for each disease.

Funding will expand the life science platform and focus on cross-disease and cross-tissue knowledge transfer, aiming for interpretable representations of disease biology with potential for personalized medicine.

A collaboration with NVIDIA aims to scale cfRNA-based training to population-scale, with Nexus as a self-supervised transformer trained on cfRNA profiles to support downstream models in oncology and other indications.

Longer-term, Network Bio envisions General Medical Intelligence that improves with more data and questions, expanding applications from biomarker discovery to diagnostics development and therapeutic target identification.

cfRNA is highlighted as an information-rich signal reflecting active gene expression across tissues in real time, enabling hundreds of millions of transcript-level observations from a single blood draw.

Industry momentum is underscored by Bristol Myers Squibb adopting NVIDIA-based AI infrastructure to boost computational power and efficiency in pharma research.

The AI architecture is designed to handle technical differences between datasets and produce interpretable insights, aiming to model biological signals at scale and explain why signals are identified.

The platform harmonizes sample selection, data quality, and formats across sites to create AI-ready, multi-modal datasets linking tissue, blood, molecular data, and longitudinal outcomes for transferable disease signals.