目的: 通过分析PubMed数据库人工智能(artificial intelligence,AI)随机对照试验(randomized controlled trials,RCTs)研究,了解AI RCTs研究现状,并为我国学者开展AI研究提供参考。 方法: 检索PubMed数据库获得AI RCTs研究。利用Web of Science获取纳入AI研究的发表时间、期刊、研究机构、国家、引文和关键词信息,采用VOSviewer 1.6.17软件对AI研究的相关信息进行提取,并对国家、研究机构和作者生成合作网络图,对关键词进行聚类分析;利用Cite Space 5.8.R3软件对被引文献和关键词共现进行突发性检测分析。 结果: 共纳入1174篇AI领域的RCTs研究,2007年后发文量增长较快;发文量最多的3个国家分别是美国、意大利和韩国,中国发文量排名第5,并与美国、加拿大和意大利等国形成了较好的合作关系;发文量最多的3个机构分别是美国哈佛大学、韩国延世大学和美国西北大学;发文量前3的作者分别是Krebs,Calabro和Picelli;高频关键词聚为3类:机器人技术在治疗中的应用、机器学习在疾病诊断和管理中的应用和AI在康复训练中的应用,结合突发性检测可以看出机器学习可能是目前的研究热点;刊文量最大的杂志是Journal of NeuroEngineering and Rehabilitation (n=40,IF2020=4.26);被引次数最高和引文爆发最强的论文都是Albert CL等人于2010年发表在The New England Journal of Medicine上的RCT,该文章分析了机器人辅助治疗中风后长期上肢损伤的效果。 结论: 开展AI RCTs研究的作者、机构和国家之间应加强合作,研究主题有待拓展。AI在医学领域具有巨大的应用前景,我国学者应加强跟踪AI在医学领域的应用趋势,加强与国外科研机构的合作,为AI的临床应用提供更多高质量的证据。
Objective: To analyze the artificial intelligence (AI) randomized controlled trial (RCT) researches published in PubMed, and explore the current situation of AI RCT researches and provide a reference for researchers to carry out related researches. Methods: AI RCTs researches were obtained by searching PubMed database, the Web of Science was used to obtain the publication year, journals, authors, organizations, and countries of included AI RCTs researches. The VOSviewer 1. 6. 17 software was used to extract the related information, and generate visual cooperation network maps for the country, organizations, citations, and authors, perform cluster analysis for keywords. The Cite Space 5. 8. R3 software was used to analyze the burst citation references and keywords. Results: A total of 1174 RCTs in AI field were included, and the number of articles increased rapidly after 2007; United States, Italy, and South Korea contributed the most publications; China ranks fifth in the number of publications and has formed good cooperative relations with the United States, Canada, and Italy; The three most contributed institutions are Harvard University of the United States, Yonsei University of Korea, and Northwestern University of the United States. Krebs, Calabro, and Picelli published the most literature; The main hot topics are the application of robot technology in treatment, the application of machine learning in disease diagnosis and management, and the application of AI in rehabilitation training. Combined with strongest citation burst, it can be seen that machine learning maybe the current research hot topic; The magazine International Journal of Neuroengineering and Rehabilitation (n=40, IF 2020=4.26) is the most productive journal; The paper with the highest number of citations and the strongest citations burst is RCT published in the New England Journal of medicine in 2010 by Albert et al, which analyzed the effect of robot-assisted treatment of long-term upper-limb injury after stroke. Conclusion: The authors, institutions, and countries of AI RCTs research should strengthen cooperation, and the research theme needs to be expanded. AI has great application prospects in the medical field, Chinese scholars should strengthen tracking the application trend of AI in the medical field, strengthen cooperation with foreign scientific research institutions, and provide more high-quality evidence for the clinical application of AI.