PALO ALTO, Calif.–Network Bio has entered into a research collaboration with Duke University to develop artificial intelligence models of human disease using tissue and blood samples linked to patient clinical histories and treatment outcomes.
The partnership will combine biospecimens from Duke University’s biobanks with molecular profiling technology and de-identified clinical records to create datasets that could help researchers identify diseases earlier, discover new drug targets and determine which treatments may be most effective for individual patients.
Under the agreement, Network Bio and Duke researchers will identify groups of patients with specific diseases and conduct molecular profiling of stored tissue and blood samples. The resulting biological data will be connected to clinical information, including diagnoses, treatments and patient outcomes, subject to applicable research protocols and privacy requirements.
The datasets will be used to train Network Bio’s AI foundation models, which are designed to examine how diseases develop and respond to treatment in humans.
Unlike many existing AI systems in biological research that rely on laboratory cell lines, animal models or publicly available datasets, the company’s approach incorporates clinical outcomes to help establish connections between molecular characteristics and disease progression.
The collaboration will also provide Duke researchers with access to newly generated molecular data, supporting additional academic studies into disease biology and treatment responses.
“Academic medical centers hold vast records of human disease: molecular data contributed by research participants, linked to their clinical outcomes,” said Asad Ali Ahmad, Co-Founder and Chief Executive Officer of Network Bio. “We are thrilled to expand our network of academic medical center biobank partners along with the University of Pennsylvania, Massachusetts General Brigham and the University of Colorado Anschutz.”
The agreement expands Network Bio’s network of academic research partners, which includes the University of Pennsylvania, Mass General Brigham and the University of Colorado Anschutz.
By integrating biological measurements with patient histories, the company aims to develop AI systems capable of identifying patterns associated with disease development, treatment effectiveness and potential therapeutic targets.
The research could also help pharmaceutical and biotechnology companies evaluate drug targets, identify biomarkers and select patient populations for clinical trials.
Svati H. Shah, M.D., Associate Dean for Translational Research at Duke University School of Medicine and Director of the Duke Center for Precision Health, said combining molecular profiling with clinical information could improve researchers’ understanding of human disease.
“Molecular data generated from cutting-edge technology together with clinical information and AI modeling has great potential for helping us better understand how diseases develop and respond to treatment,” Shah said. “This collaboration provides an opportunity to study biospecimens alongside linked clinical information, which supports deeper research that could speed discovery, while maintaining our responsibility to steward these resources carefully.”
The partnership establishes a framework for using the resulting datasets in academic research and commercial applications, including AI model development and data licensing. Those activities will be governed by the collaboration agreement and applicable data privacy and security requirements.
Raphael Potter, Co-Founder and Chief Business Officer of Network Bio, said the availability of biological samples connected to long-term patient outcomes could provide a more reliable foundation for developing predictive models of human disease.
“Biobanked tissue and blood collections, paired with years of clinical follow up, can help us train on how disease actually develops and responds to treatment in people. For our pharma and diagnostics partners, that means testing targets, biomarkers and patient selection against real human outcomes before a program ever reaches the clinic,” Potter said.
The collaboration comes as pharmaceutical companies, biotechnology developers and academic research institutions increasingly explore AI models trained on multiple types of biological and clinical data.
Such approaches are intended to move beyond isolated genetic or molecular measurements by examining how biological changes relate to disease progression and responses to treatment.
Network Bio plans to use the datasets generated through its academic collaborations to strengthen its disease modeling platform and support research into diagnostics, precision medicine and drug discovery.
