Human-Supervised, Large Language Model-Based Clinical Decision Support Aligned to National Newborn Protocols in Kenya: A Pragmatic, Early-Stage Evaluation dataset

This repository contains the datasets generated during the development and evaluation of AIFYA, a human-supervised, large language model (LLM)–based clinical decision support system (CDSS) specifically aligned with the Kenya Consolidated Newborn Clinical Protocols.The repository comprises the following three distinct datasets:Newborn Clinical Scenario Dataset: De-identified clinical parameters and management inputs corresponding to newborn patient encounters that were processed by the AIFYA platform for clinical decision support recommendations. Provides the input context for evaluating the AI system's performance.Expert Neonatal Review and Concordance Dataset: Contains the consensus ratings and detailed qualitative comments provided by independent experts in neonatal care. This dataset quantifies the inter-rater reliability and the final adjudicated agreement/concordance scores for AIFYA's recommendations against national guidelines. Supports the primary outcome analysis regarding guideline adherence and recommendation correctness.Clinician Knowledge, Attitudes, and Practices (KAP) Survey Response Data: Anonymised response data collected from the post-implementation survey administered to healthcare workers (HCWs) regarding their perceptions of AIFYA's usability, workflow integration, and the acceptance of AI in their clinical practice. Supports the secondary outcome analysis on user perception and system implementation factors.

本仓库收录了AIFYA开发与评估过程中生成的数据集。AIFYA是一款受人工监管、基于大语言模型（LLM）的临床决策支持系统（CDSS），专门适配肯尼亚《新生儿综合临床指南》。本仓库涵盖以下三个独立数据集：
1. 新生儿临床场景数据集：收录经AIFYA平台处理以生成临床决策支持建议的新生儿就诊病例的去标识化临床参数与诊疗管理输入信息，为评估该AI系统的性能提供输入上下文。
2. 新生儿专家评审与一致性数据集：收录新生儿护理领域独立专家给出的共识评级与详细定性评论。该数据集量化了AIFYA生成的建议与国家指南之间的评分者间信度，以及最终裁定的一致性/契合度得分，可支撑关于指南依从性与建议正确性的主要结局分析。
3. 临床医护人员知识、态度与实践（KAP）调查问卷应答数据：收录对医护人员（HCWs）开展的AIFYA部署后调查问卷的匿名应答数据，调研内容涵盖医护人员对AIFYA易用性、工作流程整合情况，以及临床实践中AI应用接受度的感知情况，可支撑关于用户感知与系统实施因素的次要结局分析。