# Toward an operational definition of Artificial Intelligence for health care informatics: a Delphi survey

## TL;DR

This paper presents a consensus definition of AI in healthcare to improve clarity, governance, and trust in AI systems.

## Contribution

A standardized, expert-endorsed operational definition of AI tailored for healthcare informatics.

## Key findings

- An operational definition of AI was achieved with >80% expert agreement.
- The definition emphasizes AI as a human-designed, data-driven system, avoiding anthropomorphic terms.
- The framework supports ethical governance and distinguishes AI from algorithmic tools.

## Abstract

The proliferation of Artificial Intelligence (AI) technologies, fueled by advancements in computational power and generative models, is rapidly reshaping healthcare delivery and research. However, the absence of a standardized definition of AI impedes regulatory development, confounds public discourse, and hinders clinical adoption. This study provides clarity for AI developers and users in terminology surrounding the topic, which will ultimately assist in mitigating risks to patients and the public. Utilizing a multiphase Delphi method involving international informatics experts, we synthesized existing definitions and facilitated consensus on an operational definition of AI tailored to healthcare contexts. Our findings aim to establish a foundational framework to guide ethical governance, promote funding alignment, and optimize AI integration in clinical settings.

Artificial Intelligence (AI) is transforming health care—from diagnostic imaging and predictive analytics to virtual assistants and clinical decision support. Yet a fundamental problem remains: What exactly counts as “AI”? Despite the surge of AI-enabled tools, the lack of a shared definition in healthcare informatics continues to impede policy coherence, ethical oversight, and clinical adoption. In this International Delphi Consensus Study, researchers convened informatics experts from medicine, nursing, and policy to agree on an operational definition of AI specifically for health care informatics. Participants emphasized clarity over complexity—favoring a concise description that removes anthropomorphic language such as “reasoning” or “deciding,” and highlights AI as a human-designed, data-driven system. The resulting consensus definition—endorsed by >80% of respondents—provides a common foundation for regulators, clinicians, and developers to identify what qualifies as AI in healthcare informatics. It marks a critical step toward consistent governance and public understanding, helping distinguish between algorithmic decision-support tools and truly autonomous systems. As health systems increasingly rely on AI to guide care, words matter. A clear, expert-endorsed definition is more than semantics—it's a prerequisite for transparency, accountability, and trust in the digital future of health care.