Evidence & research
Evidence and research behind governed healthcare learning
Published research and governance frameworks inform the considerations below. They do not prove Perceptors AI's effectiveness or constitute endorsement.
Ethics and oversight in healthcare AI
WHO's 2021 guidance addresses ethical challenges and governance for AI in health.
We cite it as context for accountable human oversight and careful use of AI, not as a certification of this product.
Risk management is an ongoing process
NIST's AI Risk Management Framework describes voluntary trustworthiness and risk-management considerations.
It offers a way to discuss risk. Referencing it is not a claim of NIST compliance, audit or endorsement.
Research behind source-grounded generation
Lewis and colleagues describe retrieval-augmented generation combining retrieved information with generated language.
This supports discussion of the technique. It does not demonstrate the accuracy or learning effectiveness of the Perceptors implementation.
How we review AI-generated content
The published Perceptors approach begins with approved source materials and expert knowledge.
The agent harness coordinates specialised agents with bounded permissions. Human reviewers can accept, revise, reject or return work.
People retain approval over evidence, curriculum, assessment, localisation and launch. Source versions, outputs, review decisions and releases form the documented review path.
What evidence is not available on this page
No controlled efficacy study, audited patient-outcome improvement or independently verified performance benchmark is supplied here.
A course announcement or product screenshot demonstrates what was published, not a quantified learning result.
TODO: pending verification
- TODO: Provide approved evaluations or research with methods, population, dates, denominators and limitations.
- TODO: Provide any independent review report or documented review SOP that may be published. No external audit is implied until evidence is supplied.