Why biospecimen and data sourcing is a matching problem — and how AI can help
Somewhere in the world, a freezer, a lab, or a server holds the right biospecimens or medical data that a researcher needs to answer their scientific question. Matched controls, the right disease stage, the right treatment history, the necessary longitudinal follow-up. But researchers spend months looking for what they need. And when they finally find it, the process to get access takes just as long or, worst case, doesn't work out at all.
Getting access to the right samples and datasets at the right time is a major R&D bottleneck. This is not a shortage problem, but a matching problem. Inventory and datasets are scattered across hundreds of biobanks, medical centers, and institutions, described in non-standardized ways, with varying degrees of completeness. Even when a possible option surfaces, there are still blockers, such as patient consent and terms of use, which may not align with the study aims. Meanwhile, project timelines get pushed back while samples and data that were consented for research sit unused.
BioExchange was uniquely built to solve this matching problem by leveraging AI to find the right providers for a researcher's study. Our AI-driven cataloging workflow captures deidentified structured and unstructured information shared by our providers, each with their own way of describing what they have and what they can do. We build a working model of our providers' capabilities, including biospecimen inventory, patient population and associated medical data, consent and terms of use, prospective collection capabilities, and many other attributes. Leveraging AI is what makes it possible to continually manage information shared by providers in our global network and make it usable for matching with researcher requests.
Finding high-quality providers for a researcher's project is also dependent on getting a clear and realistic understanding of what that researcher is looking for. The AI chat assistant in BioExchange was built to draw out the complete context of the researcher's requirements needed to effectively find matches. The AI matching engine then identifies the most likely set of preliminary matches based on the researcher's requirements. The researcher can then refine their requirements with the AI chat assistant, broadening or narrowing the criteria based on what is likely available for a given project.
Matching at a global scale also opens a new paradigm for researchers planning studies. Instead of just asking "Can I get this [X]?" they can now explore what studies are even possible given the capabilities across providers. Sample and data sourcing is no longer only a logistics step, but becomes something that can inform the study design itself, giving researchers confidence that a study is feasible before they invest time and resources.
Curious what's out there for your next study? Search BioExchange →
