Live AI program for doctors, 7 and 8 November. Ten seats. Reserve a seat

Sign in Join the network

29 terms
Plain language
Sourced

The vocabulary of AI in healthcare

Every term a physician, executive or researcher meets, defined once and tied to its evidence.

Search the glossary

510(k) and De NovoA 510(k) shows that a new device is substantially equivalent to a legally marketed…

Agentic AIAgentic AI is a system built around a large language model that plans, reasons.

AI hallucination in clinical contextsAn AI hallucination is output a model presents as fact that has no basis…

AlgorithmovigilanceAlgorithmovigilance is the scientific methods and activities for evaluating, monitoring, understanding.

Ambient AI scribeAn ambient AI scribe is software that listens to the clinical encounter, transcribes it.

Automation biasAutomation bias is the tendency to over-rely on an automated system's output.

Clinical decision support systemA clinical decision support system (CDSS) is software that delivers person-specific information and evidence-based…

Clinical LLMA clinical LLM is a large language model developed or adapted for healthcare tasks…

Context windowA context window is the maximum number of tokens, pieces of words.

De-identification vs anonymizationDe-identification removes or masks identifiers under a named standard; anonymization goes further.

External ValidationExternal validation is the evaluation of a predictive model on data it never saw…

Federated learning in healthcareFederated learning trains one shared model across many institutions while each keeps its own…

FHIR (Fast Healthcare Interoperability Resources)FHIR is a standard for exchanging healthcare data, published by HL7.

Fine-tuningFine-tuning is further training of an already pretrained model on a smaller.

Foundation modelA foundation model is trained on broad data at scale .

Human-in-the-loopHuman-in-the-loop is a system design in which a person reviews, approves.

Large language model (LLM)A large language model is a neural network trained on very large amounts of…

Model calibrationModel calibration is the agreement between a model's predicted probabilities and how often events…

Model cardA model card is a short document that accompanies a trained machine-learning model and…

Model Context Protocol in health ITThe Model Context Protocol (MCP) is an open standard for secure.

Model driftModel drift is the gradual decline in a deployed clinical AI model's accuracy when…

OverfittingOverfitting is when a model learns the particulars of its training data.

Positive and negative predictive value (PPV and NPV)PPV is the share of positive results that are truly positive.

Predetermined Change Control Plan (PCCP)A predetermined change control plan (PCCP) is FDA-reviewed documentation describing what modifications will be…

Prompt injection in clinical systemsPrompt injection is an attack where text hidden in an AI system's input makes…

Retrieval-augmented generation (RAG)RAG has a language model fetch relevant passages from an external knowledge source .

Sensitivity, specificity, and AUROCSensitivity and specificity describe a test at one chosen threshold.

Software as a Medical Device (SaMD)Software as a Medical Device (SaMD) is software intended for one or more medical…

Synthetic patient dataSynthetic patient data is artificially generated health records that statistically resemble real patients without…

Learn the terms here. Use them with the people in the network.

Join the network

Get the briefing

One study in healthcare AI, read closely, when it publishes.