AI-driven pre-screening for intermediate AMD
Historic pre-screen failure of roughly 60%. Eligibility signals sat in unstructured notes and imaging that no manual chart review could cover at scale.
A two-phase read: records, then imaging
A two-phase pipeline reads real-world ophthalmic records to find eligible patients for an intermediate-AMD study — de-identified, on secure infrastructure, source data resident with the site.
Historic pre-screen failure of roughly 60%. Eligibility signals sat in unstructured notes and imaging that no manual chart review could cover at scale.
- Phase 1 — generative AI reads de-identified records against inclusion and exclusion criteria
- Phase 2 — OCT image segmentation refines the shortlist against imaging criteria
- Every flagged patient carries the evidence that triggered the flag
- Deployed against data from a leading European academic eye centre
Two times the historical screen rate and pre-screen failure below 50%, per the research plan. GDPR-compliant throughout, with source data never leaving the site.
deidentify · cohort_matching · image_segmentation — a pre-screening pattern extensible to other indications and modalities.
This is the deployment behind the Clinical trials page — the same de-identify, cohort-match, and image-segmentation pattern, extensible to other indications and imaging modalities.
See the Clinical trials solution →See the other three deployments.
Client names, references, and full results available under NDA.
ECLYPSE
AI
Our proprietary AI orchestration platform for healthcare: one engine, a registry of reusable task modules and domain agents, and a governed knowledge base — deployed inside your walls and run by your team.