Evidence and research

Evidence behind governed healthcare learning

Perceptors is the agentic LMS for healthcare teams: it turns expert knowledge into reviewed courses, assessments and guided learning. This page lists the research and governance frameworks that shaped how it works, and states plainly what they do and do not show.

Ethics and oversight in healthcare AI

The World Health Organization's 2021 guidance sets out ethical principles and governance for AI in health, including human oversight and accountability. Perceptors applies that principle by keeping a named person responsible for every release.

We cite the guidance as context for our design, not as a certification or endorsement of the product.

WHO (2021): Ethics and governance of artificial intelligence for health

Risk management as an ongoing process

NIST's AI Risk Management Framework describes voluntary practices for identifying, measuring and managing AI risk across a system's life. Perceptors uses it as a shared vocabulary when institutions scope review roles and release rules.

Referencing the framework is not a claim of NIST compliance, audit or endorsement.

NIST (2023): AI Risk Management Framework

Source-grounded generation

Lewis and colleagues described retrieval-augmented generation: combining retrieved documents with generated language so that answers draw on a defined source set. Perceptors' learning guide follows the same idea and answers only from a programme's approved sources, showing where each answer comes from.

The paper supports the technique. It does not measure the accuracy or learning effect of Perceptors' implementation.

Lewis et al. (2020): Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks

How AI-drafted content is reviewed

Every programme starts from approved source materials and expert knowledge. The agent harness coordinates specialist agents with bounded permissions; reviewers can accept, revise, reject or return each piece of work.

People keep approval over evidence, curriculum, assessment, localization and launch. Source versions, agent outputs, review decisions and releases form a documented review path for each course.

Trust and governance · How the agent harness works · Course governance and accreditation

What this page does not show

No controlled efficacy study, audited patient-outcome improvement or independent performance benchmark of Perceptors is published here. A course announcement or product screenshot shows what was released, not a measured learning result.

When a programme owner approves publication of an evaluation, it will be added to this page with its method, population, dates, denominators and limitations.

Reviewing our process

Institutions evaluating Perceptors can run a partner evaluation on a real programme with agreed review criteria.

Partner evaluation

Request a briefing

See also