OnCo
ideasIdea

Send the code to the data: a federated analytics network of cancer centres

Hospitals keep their records at home; researchers send in a programme that runs at each hospital and only the summary results come back.

OHDSI has shown that a common data model (OMOP) plus federated analysis scripts can run identical studies across hundreds of databases without moving patient-level data. Oncology needs the oncology extension of OMOP (episodes, regimens, staging) completed and a standing network of 50 or more cancer centres with a governance committee that approves study packages, not data transfers. Related efforts include the EHDEN network in Europe and PCORnet in the US.

Hypothesis
A standing federated network with pre-approved study templates can answer a comparative effectiveness question on more than 10,000 patients in under six months, at least five times faster than a bespoke pooled analysis.
Rationale
OHDSI's LEGEND studies have run across tens of millions of records in months. Federated analysis avoids the legal negotiation that is the actual rate-limiting step, because no identifiable data leave the controller.
What would test it
Run three pre-registered oncology questions (for example, real-world survival on first-line immunotherapy by performance status) across at least 20 centres; measure time to result and concordance with existing trial or registry estimates.
Maturity
early clinical
Who has to act
data
Cost to try
Medium ($1M to $50M)
Years to first evidence
3
Bottlenecks it attacks
  • Data silos · Records, scans, genomes and outcomes sit in separate systems that cannot talk. Every patient's experience is lost to the next.
  • Weak real-world evidence and registries · We do not reliably know what happens to patients after approval, so we cannot tell which drugs deliver in practice.

Connected

5top