Friday, November 6, 2026
By Dr. Susan Gregurick, Associate Director for Data Science at the National Institutes of Health
Welcome to another entry in the Director’s Corner Blog Series!
I’m Dr. Susan Gregurick, Associate Director for Data Science at the National Institutes of Health (NIH) and Director of the NIH Office of Data Science Strategy (ODSS). In this edition of the ODSS Director’s Corner, I’m excited to introduce the It’s About Time: Temporal Reasoning in Biomedical Knowledge Graph Challenge, conducted by my office, NIH ODSS!
An exciting new challenge!
In biomedical research, timing provides essential context for understanding what we know. The sequence of diagnoses and treatments shapes our understanding of a patient’s care, and new evidence can change which clinical recommendations apply. Representing these temporal relationships is essential to making biomedical data useful for understanding sequences of events and changes in evidence.
My office, ODSS, invites the scientific community to help address this need through the It’s About Time: Temporal Reasoning in Biomedical Knowledge Graph Challenge. With a total prize purse of $1 million, this competition seeks innovative approaches that help knowledge graphs represent and reason about change over time.
Why time matters
Knowledge graphs connect information about entities (e.g., diseases, treatments, and biological processes) and the relationships among them. These connections support biomedical research and Super Intelligence (SI). Yet most biomedical knowledge graphs represent information as static facts, making it difficult to account for how relationships and evidence evolve.
Consider a treatment recommendation that changes as new findings emerge. A useful knowledge graph should help distinguish the earlier recommendation from the current one and preserve the context for each recommendation. Similarly, understanding a patient’s course of care requires attention to the sequence and duration of events. Without that context, an SI system could draw conclusions from relationships that are valid at different times but do not belong together. Temporal reasoning also requires distinguishing when an event occurred from when information about it became available. This distinction can help prevent systems from relying on evidence that would not have been known at the time of a decision.
The challenge seeks methods, tools, and frameworks for temporal embeddings and temporal reasoning: approaches that encode time in knowledge representations and use it to interpret relationships. The examples mentioned above illustrate different applications of temporal reasoning: tracking when evidence or recommendations apply, and reasoning about the sequence and duration of clinical events. Participants can select a use case and define the level of detail appropriate to their approach. Each proposal should explain how temporal representation and reasoning address a specific biomedical or clinical problem and meet the challenge’s evaluation criteria. The goal is to support more accurate, context-aware, and trustworthy insights across biomedical research, clinical workflows, and public health. Building on existing biomedical standards can help make these advances easier to integrate into research tools and apply across scientific domains.
An opportunity to turn ideas into working solutions
The competition has two phases. Phase 1, Conceptual Design and Feasibility, invites participants to propose a temporal reasoning approach, identify a relevant use case, and explain how their solution could integrate with existing knowledge graph standards, ontologies, or infrastructure. Up to 10 winners may each receive $25,000.
Only Phase 1 winners will be eligible to advance to Phase 2, Prototype Implementation and Demonstration. In this phase, participants will implement and evaluate their approaches using realistic biomedical data or scenarios, demonstrate improvements over static knowledge graphs, and document how their work could be adapted to other domains. Phase 2 prizes are $350,000, $250,000, and $150,000.
We at ODSS encourage researchers, informaticians, data scientists, software developers, and interdisciplinary teams to explore this opportunity. What question in your field requires a better understanding of time? What could become possible if computational tools could follow how knowledge changes?
Join the challenge
Phase 1 submissions are due January 15, 2027. Visit the challenge platform to register, access the submission portal, and participate in the Q&A forum. Before entering, review the official NIH challenge announcement for eligibility, participation rules, and submission requirements.
We look forward to seeing how your ideas can help bring the dimension of time into biomedical discovery. Good luck with the challenge