AI4HF System Requirements: A FUTURE-AI-Aligned, ISO/IEC/IEEE 29148 Requirements Dataset for Trustworthy Heart Failure Decision Support

AI4HF System Requirements Overview This dataset releases the consolidated set of 173 system requirements elicited and specified within the AI4HF project (Trustworthy Artificial Intelligence for Personalised Risk Assessment in Chronic Heart Failure) for an AI-based Clinical Decision Support System (CDSS) addressing chronic heart failure. The requirements were defined as part of Work Package 1 and consolidated in Deliverable 1.3 (November 2024). This open access release supports the project's open science commitments at M36 (May 2026) and is intended to enable reuse by researchers, developers, regulators, and HTA practitioners working on trustworthy clinical AI. Scope and methodology The specification operationalises high-level trustworthy AI principles into engineering-grade requirements through two complementary instruments.Normative coverage is provided by the FUTURE-AI framework (Lekadir et al., BMJ 2025). Each requirement is mapped to a sub-principle of its taxonomy, which is organised around six guiding principles (i.e., Fairness, Universality, Traceability, Usability, Robustness, and Explainability) and a set of cross-cutting General recommendations applicable across the AI lifecycle. Syntactic and structural rigour is provided by the ISO/IEC/IEEE 29148:2011 standard on requirements engineering. Each requirement statement follows the prescribed natural language template and is classified as either functional (FR) or non-functional (NFR). Requirements were elicited from heterogeneous stakeholder workstreams including patients, healthcare professionals, ethical, legal and societal experts, health technology assessment experts, and AI Product Passport developers. Additional inputs were drawn from foundational project documents such as the Description of Action and the Grant Agreement. All requirements were subsequently harmonised, deduplicated, and verified for traceability against the FUTURE AI principles. File contents The dataset is distributed as a single workbook containing two sheets. Requirements This sheet contains the 173 system requirements, one requirement per row, with the following fields: ID, System requirement, Rationale, Stakeholder(s) affected, FUTURE-AI principle, Source, Type, and Means of verification (MoV). Codebook This companion sheet documents the meaning of each column, the interpretation of the Source field, the complete acronym glossary, and the FUTURE AI principle taxonomy applied throughout the dataset. Related publications Lekadir, K. et al. (2025). FUTURE AI: international consensus guideline for trustworthy and deployable artificial intelligence in healthcare https://doi.org/10.1136/bmj-2024-081554 AI4HF (https://www.ai4hf.com) Funding This work was supported by the European Union's Horizon Europe research and innovation programme under the AI4HF project (Grant Agreement No. 101080430).AI4HF 系统需求 ## 概述 本数据集发布了AI4HF项目（慢性心力衰竭个性化风险评估可信赖人工智能项目）中征集并梳理的共计173项系统需求，该项目旨在开发面向慢性心力衰竭的人工智能临床决策支持系统（Clinical Decision Support System，CDSS）。相关需求作为工作包1的组成部分，于2024年11月发布的可交付成果1.3中完成整合。本次开源发布契合项目在第36个月节点（2026年5月）的开放科学承诺，旨在为致力于可信赖临床人工智能研究的科研人员、开发者、监管机构以及卫生技术评估（Health Technology Assessment，HTA）从业者提供复用支持。 ## 范围与方法学 本规范通过两类互补工具，将高层次可信赖人工智能原则转化为工程级需求。规范性覆盖依托FUTURE-AI框架（FUTURE-AI framework，Lekadir等，《英国医学杂志》2025年）完成。每项需求均映射至其分类体系的子原则，该体系围绕六大指导原则（公平性、普适性、可追溯性、可用性、鲁棒性与可解释性）以及一组适用于人工智能全生命周期的跨领域通用建议构建。 本规范的语法与结构严谨性遵循ISO/IEC/IEEE 29148:2011《需求工程》标准。每项需求陈述均遵循规定的自然语言模板，并被归类为功能需求（FR）或非功能需求（NFR）。 需求征集自多元利益相关方工作流，涵盖患者、医护人员、伦理法律与社会专家、卫生技术评估专家以及人工智能产品护照开发者。额外输入源自项目核心文件，如行动说明书与资助协议。所有需求后续均经过统一协调、去重处理，并针对FUTURE-AI原则完成可追溯性验证。 ## 文件内容 本数据集以单个工作簿形式分发，包含两个工作表。 ### 需求工作表 该工作表包含173项系统需求，每行对应一项需求，字段包括：ID、系统需求描述、合理性依据、受影响利益相关方、FUTURE-AI原则、来源、类型以及验证手段（MoV）。 ### 代码手册工作表 该配套工作表说明各列含义、来源字段的解读规则、完整缩写词表以及本数据集所采用的FUTURE-AI原则分类体系。 ## 相关出版物 Lekadir K 等 (2025). FUTURE AI：医疗领域可信赖且可部署人工智能的国际共识指南 https://doi.org/10.1136/bmj-2024-081554 AI4HF项目官网：https://www.ai4hf.com ## 资助信息 本研究获欧盟“地平线欧洲”研究与创新计划资助，资助项目为AI4HF项目（资助协议编号：101080430）。