Glossary
The vocabulary of AI in healthcare, in plain language
Every term a clinician or health leader meets in an AI conversation, defined clearly and tied to the evidence behind it.
Agentic AI
What agentic AI means in healthcare — the working definition from the peer-reviewed literature, how these systems differ from chatbots, and what benchmarks show they can and cannot yet do. As of July 2026.
TermAI hallucination in clinical contexts
What it means when a clinical AI system fabricates content — the measured rates, the severity splits, and the failure modes specific to healthcare. As of July 2026.
TermAlgorithmovigilance
The discipline of watching clinical algorithms after deployment — where the term came from, the failures that made it necessary, and the systems and rules now operationalizing it. As of July 2026.
TermAmbient AI scribe
What an ambient AI scribe is, how the capture-to-signed-note pipeline works, and what the largest deployments have measured — time saved, burnout moved, and the failure modes that keep clinicians reviewing every note. As of July 2026.
TermClinical decision support system
What a clinical decision support system is, what the pooled trial evidence shows it achieves, and where the FDA device boundary now sits after the January 2026 guidance revision. As of July 2026.
TermClinical LLM
What a clinical LLM is, how general models are adapted for healthcare work, what benchmark scores do and do not prove, and where regulators have drawn the lines. As of July 2026.
TermDe-identification vs anonymization
Two words the healthcare data world uses interchangeably and shouldn't — what separates de-identified data from truly anonymous data under US and EU law, and why the gap between them is where re-identification risk lives. As of September 2026.
TermExternal Validation
What it means to test a clinical prediction model on data it has never seen — how rarely it happens, how often performance drops, and why the distinction decides whether AI can be trusted at your hospital. As of July 2026.
TermSoftware as a Medical Device (SaMD)
What counts as Software as a Medical Device, where the regulatory boundary runs, and how the FDA handles AI products that carry the label. As of July 2026.
TermFederated learning in healthcare
What federated learning means when the data is patient data — how institutions train a shared model without moving records, what it has already delivered across dozens of hospitals, and why it is a privacy tool rather than a privacy guarantee. As of September 2026.
TermFoundation model
What a foundation model is, where the term came from, how the idea landed in healthcare — from retinal imaging to pathology — and how the EU AI Act and WHO define and govern the class. As of July 2026.
TermHuman-in-the-loop
What human-in-the-loop means in healthcare AI — the regulatory anchors that make it mandatory for high-stakes systems, and the automation-bias evidence that shows why nominal oversight can fail. As of July 2026.
TermModel calibration
Whether a clinical AI model's stated probabilities can be trusted as probabilities — what calibration means, why it decides safe use, and how it differs from a model's ranking ability. As of July 2026.
TermModel Context Protocol in health IT
The open standard that lets AI models plug into electronic health records through one interface instead of a hundred custom integrations — what it is, where it already touches FHIR data, and the security rules that come with connecting a model to a live record. As of September 2026.
TermModel drift
When a clinical AI model quietly gets worse because the world it was trained on has moved on — what model drift is, why healthcare feels it acutely, and how it is monitored. As of July 2026.
TermPredetermined Change Control Plan (PCCP)
How the FDA pre-authorizes future updates to AI-enabled medical devices — what a PCCP contains, where the idea came from, and how many devices actually carry one. As of July 2026.
TermPrompt injection in clinical systems
The security flaw at the top of the OWASP LLM risk list, explained for healthcare — where the malicious text can hide in a scanned slide, a pasted referral, or a record field, and where two published studies already show it changing model output. As of September 2026.
TermRetrieval-augmented generation (RAG)
What retrieval-augmented generation means in healthcare — how grounding a language model in curated clinical sources changes its answers, what the measured gains are, and the failure modes retrieval introduces. As of July 2026.
TermSensitivity, specificity, and AUROC
Two ways to describe how well a clinical test or model separates sick from well — one at a single decision point, one across all of them — and why the headline number can hide the one that matters. As of July 2026.
TermSynthetic patient data
Artificial health records that resemble real patients without being any of them — what synthetic patient data is, where it genuinely helps, and why "synthetic" and "private" are not the same claim. As of July 2026.