Alboksmaty Ahmed, Aldakhil Reham, Hayhoe Benedict W J, Ashrafian Hutan, Darzi Ara, Neves Ana-Luisa
Institute of Global Health Innovation (IGHI), Department of Surgery and Cancer, Imperial College London, St Mary's Campus, Norfolk Place, London W2 1PG, UK.
Department of Primary Care and Public Health, School of Public Health, Imperial College London, 90 Wood Ln, London W12 0BZ, UK.
EBioMedicine. 2025 Jul 21;118:105861. doi: 10.1016/j.ebiom.2025.105861.
BACKGROUND: AI-powered Voice-to-text Technology (AIVT) offers a promising solution to reduce clinicians' documentation burden during consultations, allowing more focus on patient interaction. This systematic review assesses AIVT's impact on care quality in primary care and outpatient settings, focusing on seven components: effectiveness, efficiency, safety, patient-centredness, timeliness, equity, and integration. METHODS: A systematic search of five databases (Medline, Embase, Global Health, CINHAL, Scopus) was conducted for studies published up to September 20, 2024. Studies were included if they assessed the use of AIVT for medical documentation in primary care or outpatient settings, compared to manual or non-AI documentation methods, and reported outcomes relevant to the seven quality components. A narrative synthesis was conducted; meta-analysis was unfeasible due to study heterogeneity. FINDINGS: Of 1924 papers, nine studies were included (n = 524 healthcare professionals, n = 616 patients, 1069 consultations). Most (n = 7) were from the USA, with others in Bangladesh and the Philippines. All studies assessing effectiveness, patient-centredness, and efficiency (n = 9, 6, and 5, respectively) reported improvements, including faster documentation, reduced administrative burden, and enhanced patient-provider interaction. Safety findings were inconclusive; three of six studies raised concerns. Four studies highlighted seamless AIVT integration with Electronic Health Records, improving service timeliness. Three studies identified equity issues, referring to limited diversity and controlled simulation settings. INTERPRETATION: AIVT tools enhance documentation efficiency and patient-centred care, but concerns over transcription errors and generalisability warrant further testing in large-scale, diverse real-world settings. FUNDING: This study was supported by the National Institute for Health and Care Research (NIHR) North-West London Patient Safety Research Collaboration (NIHR NWL PSRC, Ref. NIHR204292), with infrastructure support from the NIHR Imperial Biomedical Research Centre.
背景：人工智能语音转文本技术（AIVT）为减轻临床医生在会诊期间的文档记录负担提供了一个有前景的解决方案，使他们能够更加专注于与患者的互动。本系统评价评估了AIVT对初级保健和门诊环境中医疗质量的影响，重点关注七个方面：有效性、效率、安全性、以患者为中心、及时性、公平性和整合性。 方法：对五个数据库（Medline、Embase、Global Health、CINHAL、Scopus）进行系统检索，查找截至2024年9月20日发表的研究。如果研究评估了AIVT在初级保健或门诊环境中用于医疗文档记录的情况，与手动或非人工智能文档记录方法进行了比较，并报告了与七个质量方面相关的结果，则纳入该研究。进行了叙述性综合分析；由于研究的异质性，无法进行荟萃分析。 结果：在1924篇论文中，纳入了9项研究（524名医疗保健专业人员，616名患者，1069次会诊）。大多数研究（7项）来自美国，其他研究来自孟加拉国和菲律宾。所有评估有效性、以患者为中心和效率的研究（分别为9项、6项和5项）均报告有改善，包括文档记录更快、行政负担减轻以及患者与提供者之间的互动增强。安全性结果尚无定论；六项研究中的三项提出了担忧。四项研究强调了AIVT与电子健康记录的无缝整合，提高了服务及时性。三项研究发现了公平性问题，指出多样性有限且模拟环境受到控制。 解读：AIVT工具提高了文档记录效率和以患者为中心的护理，但对转录错误和普遍性的担忧需要在大规模、多样化的现实环境中进一步测试。 资金：本研究由国家卫生与保健研究所（NIHR）伦敦西北部患者安全研究合作项目（NIHR NWL PSRC，编号NIHR204292）资助，并得到了NIHR帝国生物医学研究中心的基础设施支持。
JBI Database System Rev Implement Rep. 2016-4
JMIR Med Inform. 2025-8-1
Cochrane Database Syst Rev. 2022-5-20
Health Technol Assess. 2001
BMC Med Inform Decis Mak. 2025-7-1
Health Soc Care Deliv Res. 2025-5-21
Health Technol Assess. 2024-10
JMIR AI. 2024-10-4
Prev Chronic Dis. 2024-8-22
AMIA Jt Summits Transl Sci Proc. 2024-5-31
J Am Med Inform Assoc. 2024-4-3
BMC Prim Care. 2023-11-3
BMC Med Educ. 2023-9-22
J Pers Med. 2023-6-5
Zotero 插件