Deep 6 AI vs xCures
Both read records that were never designed to be machine read, and they start from different failures. Deep 6 reasons over the unstructured chart an institution already holds, matching patients and sites in real time across therapeutic areas and covering feasibility as well as enrolment. xCures assembles the chart in the first place, retrieving across local, state and national exchanges, publishing how each extractor is validated and stating plainly that it acts as its customer's business associate. A cancer centre or research site whose patients receive care elsewhere has the xCures problem before it has the Deep 6 one. Neither publishes how often an eligible patient never surfaces, which remains the only measure that would settle either product's value.
- It reads unstructured clinical data across notes, pathology, genomics and laboratory reports to match patients and sites, and covers the trial lifecycle from feasibility through enrolment.
- It works across therapeutic areas rather than one specialty, so a health system running trials in several services gets one platform.
- It quantifies how much decisive eligibility information sits in unstructured data, which is the argument for reading notes rather than filtering codes.
- Retrieval across local, state and national exchanges is the core competency, assembling histories for patients whose care happened somewhere else entirely.
- It states without qualification that it operates as its customer's business associate and publishes how each extractor is validated.
- Treatment option outputs pass through a defined review step rather than reaching a clinician raw.
These sit at adjacent layers and are often complementary, since one reasons over the chart an institution holds and the other assembles the chart it does not. Neither publishes a false negative rate for eligibility screening, which is the decisive measure because an eligible patient who never surfaced is invisible in every metric either reports. Deep 6 publishes no attestation scoped to the platform, no business associate terms and no governance framework. Both inherit documentation bias, so patients with sparse records are systematically less likely to be found.