By Eric Andrew Kristof, RN, CAHIMS | Healthcare IT Professional | HIMSS Arkansas Chapter Member
About This Series: The HIMSS Word of the Day is a running blog series drawing from the HIMSS Dictionary of Health Information and Technology Terms, Acronyms, and Organizations, 6th Edition (2025) — the authoritative reference for healthcare IT professionals and the primary study resource for CAHIMS and CPHIMS certification. Each post paraphrases the term in plain language, grounds it in real-world healthcare IT context, and adds a bedside nursing perspective you won’t find in a textbook.
📖 The Term: Artificial Intelligence (AI)
Source: HIMSS Dictionary of Health Information and Technology Terms, Acronyms, and Organizations, 6th Edition (2025)
🔍 Plain-Language Definition
Artificial Intelligence (AI) in healthcare is the theory and development of computer systems capable of performing tasks that historically required human intelligence — including visual perception, speech recognition, pattern detection, natural language understanding, decision-making, and translation between languages.
That definition, from the HIMSS Dictionary, is deliberately broad. In practice, “healthcare AI” is not a single technology but a family of techniques that share the goal of computationally augmenting or automating cognitive work.
Four categories of AI dominate current healthcare use cases:
| Category | What It Does | Healthcare Examples |
|---|---|---|
| Rule-Based Systems | Follow explicit logical rules authored by humans | Traditional Clinical Decision Support System (CDSS) alerts, drug-interaction checkers, order sets |
| Machine Learning (ML) / Predictive Analytics | Learn statistical patterns from training data | Sepsis early-warning scores, readmission risk models, imaging diagnostic aids |
| Natural Language Processing (NLP) | Extract structured meaning from unstructured text and speech | Ambient scribes, chart abstraction, medical coding assistance, symptom triage chatbots |
| Generative AI / Large Language Models (LLMs) | Generate new text, images, or code from learned patterns | Draft responses to patient messages, note generation, education content, coding assistance |
⚠️ Exam Alert: AI is not a monolith on the CAHIMS exam. Distinguish rule-based systems (deterministic, explicit logic — traditional CDSS) from machine learning (statistical pattern recognition from training data), from generative AI (produces new content). Also know the difference between AI (the broader field) and ML (a subset), and between narrow AI (task-specific, everything in current healthcare use) and artificial general intelligence (hypothetical, not a real healthcare product category). Answer choices that use “AI” and “machine learning” as synonyms are usually the wrong answer.
🌐 Why It Matters: Real-World Healthcare IT Context
AI has moved from research demo to production infrastructure in healthcare over the past five years, and the regulatory environment is racing to keep up.
The ONC Health Data, Technology, and Interoperability (HTI-1) Final Rule, issued in 2024, established transparency requirements for AI and predictive decision support integrated with certified EHR technology. Certified EHR vendors must now provide information about the algorithms, training data, and intended use of any predictive AI they incorporate — a significant policy step toward the “AI nutrition label” concept.
The FDA regulates AI-enabled medical devices under its Software as a Medical Device (SaMD) framework, with a growing list of authorized AI-enabled devices — over 1,000 by 2025, most of them in radiology. FDA published its AI/ML Software as a Medical Device Action Plan and its Good Machine Learning Practice (GMLP) guiding principles in partnership with Health Canada and MHRA.
The WHO issued global guidance on AI ethics in health, emphasizing autonomy, transparency, inclusiveness, accountability, and sustainability as core principles for responsible healthcare AI deployment.
Healthcare AI failure modes and risks:
- Bias inherited from training data — models trained on nonrepresentative populations perform worse for underrepresented groups
- Distributional shift — models degrade when the population or workflow differs from training conditions
- Automation bias — clinicians over-rely on AI outputs and disengage clinical judgment
- Alert fatigue extended into AI-driven predictions
- Privacy risks from large training data sets containing PHI
- Hallucinations in generative AI producing confident, factually incorrect clinical content
- Regulatory uncertainty for AI that adapts after deployment
Strengths of mature healthcare AI governance:
- Clearly defined intended use and validation for the local patient population
- Ongoing performance monitoring against operational metrics
- Transparent model documentation (“AI nutrition labels”)
- Multi-disciplinary AI governance committees including clinicians, informatics, compliance, security, ethics, and equity
- Human oversight built into workflow — AI recommends; clinicians decide
- Formal bias auditing across demographic subgroups
- Continuous training of the workforce on AI capabilities, limitations, and appropriate reliance
The organizations getting AI right are treating it as a governance discipline, not a shopping list of point solutions.
🏥 The Nurse’s Perspective: Clinical, Bedside, & Workflow
Nurses are already working alongside AI every day, whether they recognize it or not.
At the Bedside
Sepsis early-warning scores that fire in the chart are AI. Fall risk models that update with each Braden score entry are AI. Chatbots that route patient messages to the right care team are AI. Ambient documentation tools that draft the visit note from the audio of the encounter are AI. Predictive staffing models that determine unit ratios are AI.
Nurses experience AI most acutely when it works well and disappears into workflow, and most acutely when it fails and creates work: a sepsis alert that fires on the third day of a well-managed septic patient, an ambient scribe that misheard the diagnosis, a predicted-discharge model that plans transportation for a patient who is still febrile.
The nurse’s judgment remains the safety layer. AI predictions are inputs; nursing assessment, clinical reasoning, and patient advocacy are the decision layer.
In Clinical Workflow
Nursing AI workflows increasingly span shift assignment, patient acuity classification, admission and discharge prediction, early-warning surveillance, medication administration decision support, and documentation assistance. Ambient AI scribes are moving from provider workflows into nursing documentation, with early evidence of documentation time reductions.
The persistent risk is automation bias — the tendency to trust AI outputs without adequate scrutiny. Nurses are often taught to trust the technology because it is presented as objective. Nursing informatics has an increasing responsibility to train and equip the nursing workforce to critically evaluate AI outputs, question outliers, and maintain diagnostic skepticism.
In EHR Implementation (Go-Live Perspective)
During Epic Go-Lives that include AI features (sepsis models, ambient scribes, MyChart AI drafts, deterioration index), the build-time decisions are consequential. Which populations does the model apply to? What is the alert threshold and workflow? Who is the accountable clinician for each recommendation? What is the escalation pathway when AI outputs conflict with clinical judgment? Where does documentation of the AI recommendation and the clinical response live?
End-user support analysts during AI-enabled Go-Lives face a novel challenge: users often cannot tell whether a surprising output is a bug, a training gap, a model limitation, or an appropriate flag. Structured feedback pathways to the AI governance committee, distinct from routine break-fix tickets, are essential. Not every “the alert seems wrong” report is the same issue — some are model calibration problems, some are workflow design issues, some are individual training gaps, and some are legitimate clinical judgment differences.
Understanding that AI in healthcare is a sociotechnical system — where the technology, the workflow, the clinician, and the governance all interact — is the defining competency of an AI-fluent informatics professional.
🎓 CAHIMS / CPHIMS & HIMSS Perspective
CAHIMS Exam Domain Mapping
AI appears across CAHIMS exam domains, with rapidly growing weight:
| CAHIMS Domain | How AI Appears |
|---|---|
| Clinical Informatics | The primary domain; AI-driven CDSS; predictive analytics; ambient documentation; automation bias; human-in-the-loop workflow design |
| Healthcare & Technology Environments | ONC HTI-1 transparency requirements; FDA AI/ML SaMD framework; WHO ethics guidance; state and federal AI legislation |
| Healthcare Information & Systems Management | Data quality prerequisites for AI; model deployment and monitoring; MLOps; integration architecture |
| Management & Leadership | AI governance committee structure; ethics and equity oversight; workforce training; procurement and vendor evaluation |
Key Terms to Know Alongside AI
- Machine Learning (ML) — a subset of AI focused on statistical pattern recognition from training data; not synonymous with AI
- Natural Language Processing (NLP) — the AI subdomain concerned with understanding and generating human language
- Generative AI / Large Language Models (LLMs) — AI systems that produce new content (text, images, code) from learned patterns; distinct from predictive AI
- Automation Bias — the human tendency to over-rely on automated system outputs; a primary healthcare AI safety concern
- Model Bias / Algorithmic Fairness — systematic performance differences across demographic subgroups, often from unrepresentative training data
- AI Governance — the interdisciplinary framework for evaluating, deploying, monitoring, and retiring AI in healthcare organizations
HIMSS Organizational Position
HIMSS supports responsible AI adoption through public policy engagement with federal agencies, research from HIMSS Global Insights, and AI-focused programming at its Global Health Conference. HIMSS members can find AI governance guidance, case studies, and research through the HIMSS Resource Center.
🔗 Explore Further: External References
- ONC — HTI-1 Final Rule and AI Transparency https://healthit.gov/regulations/hti-rules/hti-1-final-rule/
- FDA — Artificial Intelligence-Enabled Medical Devices https://www.fda.gov/medical-devices/digital-health-center-excellence/artificial-intelligence-enabled-medical-devices
- FDA — Good Machine Learning Practice (GMLP) Guiding Principles https://www.fda.gov/medical-devices/artificial-intelligence-enabled-medical-devices/good-machine-learning-practice-medical-device-development-guiding-principles
- World Health Organization — Ethics and Governance of Artificial Intelligence for Health https://www.who.int/publications/i/item/9789240029200
- AMIA — Artificial Intelligence in Healthcare https://amia.org/community/topics/artificial-intelligence-healthcare
- HIMSS — Resource Center https://www.himss.org/resource-center
Term source: HIMSS Dictionary of Health Information and Technology Terms, Acronyms, and Organizations, 6th Edition (2025). All definitions in this series are paraphrased for editorial purposes. Readers are encouraged to consult the primary source for exact language.
Eric Andrew Kristof, RN, CAHIMS is a Healthcare IT professional and HIMSS Arkansas Chapter member based in Hot Springs Village, AR, with hands-on Epic Go-Live experience and a diverse healthcare and IT background. He writes at kristof.org at the intersection of clinical care and healthcare technology.
Leave a Reply