Oracle and AstraZeneca announced a non-binding agreement on 20 October to explore AI-enabled links between clinical research and routine care. The proposed work includes finding potentially eligible trial participants, generating evidence from everyday health data and bringing research insights back into treatment. No recruitment or patient-outcome results accompany the announcement.[1]
The distinction between data and evidence is central to that plan. The FDA describes real-world data as information routinely collected about health and healthcare, including medical records, insurance claims and disease registries. Evidence emerges from analysing those records to address a product’s use, potential benefits or risks. A larger collection of records is therefore an input to research; the study question and analysis determine what can be learned from it.[2]
For example, a hypothetical clinic might have records showing which patients received a medicine and which later returned with symptoms. Those entries could help identify a research population. Establishing whether the medicine caused the difference requires an appropriate comparison, including consideration of why different patients received different treatments. Connecting databases makes the question easier to investigate without resolving it automatically.[2]
There is a similar handover in recruitment. NIH guidance describes pre-screening with a study team, an informed-consent process and a fuller review of medical history and any necessary tests. A promising match in a health record opens a route to that conversation. It does not itself enrol the patient.[3]
The companies envisage shared governance, reusable technical components and agreed measures of progress. Their initial therapeutic interests include cancer, cardiovascular and metabolic conditions, respiratory disease and rare diseases.[1]
Our test would follow each handover: whether a suggested participant reaches useful screening, whether an analysis changes a study decision, and whether a finding reaches the care team in time. Measuring these outcomes separately would show where the connection actually helps.
Sources & further reading
- Oracle and AstraZeneca announce AI healthcare collaboration Oracle · 20 Oct 2026 · Accessed 20 Oct 2026
- Real-World Evidence US Food and Drug Administration · Accessed 20 Oct 2026
- What to Expect During a Clinical Trial National Institutes of Health · Accessed 20 Oct 2026