Researchers introduced DeepEvidence to streamline scientific discovery
The new agent moves beyond basic fact-finding to synthesize and map complex scientific evidence for drug research.
Researchers from AstraZeneca and Turbine Simulated Cell Systems published a paper in Nature Machine Intelligence on October 1, 2026, detailing a new research agent called DeepEvidence.
Traditional AI tools often focus on retrieving existing documents or facts. DeepEvidence takes a different approach by building explicit, structured representations of scientific findings.
DeepEvidence, a new deep research agent, goes beyond retrieving facts and constructs explicit representations of scientific evidence.
In fields like oncology, the challenge is no longer a lack of data. Instead, the bottleneck is the ability to connect disparate pieces of information to identify new targets for medicine.
DeepEvidence works by mapping relationships between scientific evidence. Think of it as a librarian who not only finds the books you asked for but also draws a map of how the ideas inside them relate to one another.
This system helps scientists move from simple searching to active synthesis. It integrates data to provide a clearer picture of biological systems.
Would you trust a machine to connect the dots in a complex medical research study?
Because this system requires integration with specific scientific knowledge graphs and datasets, it is not yet a general-purpose tool available for public testing. The paper focuses on its application in high-level biomedical discovery rather than broad consumer research.
Source: Nature Machine Intelligence