Human tissue, longitudinal outcomes and multimodal data underpin a new platform seeking transferable signals across disease.
Network Bio has launched with $50 million in financing to expand an AI platform built around large-scale human biological datasets, combining patient-derived tissue and paired blood samples with molecular information and longitudinal clinical outcomes. Investors include Section 32, Thiel Bio, Founders Fund, Breyer Capital, Blue Venture Fund and JSL Health Capital.
The Palo Alto biotech is developing disease-specific models intended to support personalized medicine, drawing on a research network of academic biobanks at institutions including Mass General Brigham, the University of Pennsylvania and the University of Colorado Anschutz. Although Network Bio is not explicitly a longevity company, its effort to interrogate shared biology across metabolic, cardiovascular, immune and other diseases puts its platform close to a central question in geroscience: what can be learned when chronic diseases are considered not simply as separate endpoints, but as overlapping manifestations of biology changing over time?
Longevity.Technology: Network Bio may not describe itself as a geroscience company, but there is something distinctly geroscientific about refusing to let human biology remain tidily divided into disease-sized boxes. Aging is, after all, an enthusiastic breaker of silos – metabolic dysfunction, cardiovascular disease, immune dysregulation and other chronic conditions accumulate and interact across decades, often sharing biology long before they acquire separate diagnostic labels. The intriguing part of Network Bio’s proposition is therefore less that AI can find patterns in very large datasets – by now, that particular rabbit has been pulled from the hat rather often – and more that those datasets combine patient tissue, blood, molecular information and longitudinal clinical outcomes, potentially allowing models to ask which biological signals persist across tissues, diseases and time. For geroscience, that matters: much of the field’s promise rests on identifying processes that precede and contribute to multiple age-related diseases, then intervening early enough to alter trajectory rather than merely treating the consequences. If Network Bio can distinguish genuinely transferable biology from correlations produced by scale, heterogeneity and technical noise, such models could help close one of longevity research’s more stubborn gaps – the distance between mechanistic insight and meaningful human prevention. The important word, of course, is if; a large biological atlas can provide an exceptionally detailed map, but healthcare only changes when we discover which roads are worth taking.
Building the biological substrate
Network Bio’s research network is designed to address a familiar problem in biomedical AI: data may be abundant within individual institutions, but differences in collection, processing and annotation make combining it considerably less straightforward. That matters more in aging research than most fields, where heterogeneity – divergent trajectories with age, exposure, disease burden and treatment history – is itself the signal, not noise to be averaged away.
“Every patient leaves a barcode of their disease in their tissue, and until now no one has been able to read those barcodes at scale,” said Asad Ali Ahmad, PhD, CEO and Cofounder of Network Bio. “Working with the biobanks of some of the country’s leading academic medical centers, we built the network that makes reading them possible. We then built an AI platform that is already finding signals in conditions it was never trained on. This is the future of medicine.”
Learning across disease
Rather than designing a separate computational system for each indication, Network Bio says its architecture is intended to learn biological principles capable of transferring across diseases, tissues and data modalities. Its platform combines the biobank network with what it describes as bio-native AI, designed to account for technical confounders and generate interpretable representations of disease biology. The company points to peer-reviewed publication in journals including Nature Machine Intelligence as evidence the approach holds up outside its own walls; published work has included respiratory disease, with additional studies in ovarian and bone disease.
That ambition is shared beyond the founding team. “What differentiates Network Bio is the combination of the network of academic medical center biobanks with an AI architecture built for medicine,” said Mike Pellini, Managing Partner at Section 32 and Chairman of Network Bio’s board. “Rather than building models one disease at a time, the company is building General Medical Intelligence that becomes more capable with every new biological question it answers.” It’s a tidy phrase for the underlying claim – but the claim itself still needs the same scrutiny as any other; generalization must survive independent populations and clinical scrutiny before it earns the name.
Network Bio has already secured a co-development and licensing collaboration worth more than $30 million with an unnamed Fortune 100, top-10 healthcare company. Commercial interest of that scale is a meaningful vote of confidence – though it is not itself clinical validation; that test still lies ahead, in whether the models’ signals ultimately improve diagnostics, therapeutic development or patient stratification.
Cofounders Hani Goodarzi, PhD and Raphael Potter frame the challenge as one of pairing access with interpretation. “We built Network Bio based on the belief that AI and data hold the keys to advancing the human condition. We have been focused tirelessly on the dual mandate of gaining access to the tissue datasets which explain disease and building the AI systems capable of interpreting the complex results,” they said.
From signals to trajectories
For healthy longevity, the longer-term opportunity lies upstream. If multimodal models can connect molecular changes not merely with disease status but with subsequent clinical trajectories, they may help identify when ostensibly healthy biology begins edging toward pathology – and where prevention might usefully intervene.
That remains a considerable translational journey. But as geroscience increasingly turns from cataloging mechanisms toward changing outcomes in humans, better maps of how biology unfolds across time, tissues and diseases may prove less an accessory to the journey than part of the road itself.
