Chitrapady Shravya, Rajendran Rajalakshmi, K Haritha, M U Tejashree, Rashid Muhammed, Poojari Pooja Gopal, K Vijayanarayana, Varma Muralidhar, Devi Vasudha, Acharya Dinesh, Khan Sohil, Thunga Girish
Department of Pharmacy Practice, Manipal College of Pharmaceutical Sciences, Manipal Academy of Higher Education, Manipal, Karnataka, India.
Department of Pharmacotherapy, University of Utah, Salt Lake City, Utah, USA.
BMJ Open. 2025 Aug 25;15(8):e102158. doi: 10.1136/bmjopen-2025-102158.
INTRODUCTION: Recent advancements in diagnosing tropical fevers increasingly use artificial intelligence (AI). These innovations focus on diagnosing single or multiple diseases, significantly reducing the global burden of tropical fevers. This protocol helps to identify the key factors required for a systematic review of AI-based machine learning (ML) diagnostic test accuracy-based studies to obtain a view on the pooled performance of different types of available tools. This systematic review protocol aims to review the type of ML-based AI tools and pool the performance metrics of the currently available ML-based AI devices. METHODS AND ANALYSIS: Patients with tropical fevers will be recruited, whereas the ML-based AI model will be the index test, and dengue, scrub typhus, leptospirosis, malaria, influenza, typhoid, chikungunya and Japanese encephalitis will be the target conditions considered for the review. Search does not restrict to any time period, and all original research studies with cross-sectional study design that are related to the development of ML tools or specific algorithms used for the diagnosis of tropical fevers from the date of inception until the date will be considered for review.Specific keywords and relevant MeSH terms for 'artificial intelligence', 'diagnosis, and 'tropical fevers' will be selected. A systematic search will be conducted in Medline/PubMed, Embase, Cochrane and Scopus covering literature from inception to February 2025. Upon retrieval of all the studies into an Excel sheet, deduplication will be done, followed by initial and secondary screening. Data extraction will be conducted using Microsoft Excel. The obtained data will be summarised narratively, and a meta-analysis of quantitative data will be performed using Meta-Disc software. The Quality Assessment of Diagnostic Accuracy Studies 2 tool will be employed to evaluate the quality of the studies. The study is planned to start in March 2025 and will be completed by September 2025. ETHICS AND DISSEMINATION: Ethical approval is not required for this systematic review and meta-analysis, as it will use data from previously published studies. The results of the review will be published in academic journals and presented at international conferences. PROSPERO REGISTRATION NUMBER: CRD42024516128.
引言：热带发热诊断领域的最新进展越来越多地采用人工智能（AI）。这些创新聚焦于诊断单一或多种疾病，显著减轻了热带发热的全球负担。本方案有助于确定对基于人工智能的机器学习（ML）诊断测试准确性研究进行系统评价所需的关键因素，以便了解不同类型现有工具的综合性能。本系统评价方案旨在回顾基于ML的AI工具类型，并汇总当前可用的基于ML的AI设备的性能指标。 方法与分析：将招募热带发热患者，以基于ML的AI模型作为指标测试，登革热、恙虫病、钩端螺旋体病、疟疾、流感、伤寒、基孔肯雅热和日本脑炎作为本次评价考虑的目标疾病。检索不限任何时间段，从开始日期至当前日期，所有采用横断面研究设计且与用于热带发热诊断的ML工具或特定算法开发相关的原始研究均将纳入评价。将选择“人工智能”“诊断”和“热带发热”的特定关键词及相关医学主题词（MeSH）。将在Medline/PubMed、Embase、Cochrane和Scopus中进行系统检索，涵盖从开始到2025年2月的文献。将所有检索到的研究录入Excel表格进行重复数据删除，随后进行初筛和二次筛选。将使用Microsoft Excel进行数据提取。所获数据将进行描述性总结，并使用Meta-Disc软件对定量数据进行Meta分析。将采用诊断准确性研究质量评价2工具评估研究质量。本研究计划于2025年3月开始，2025年9月完成。 伦理与传播：本系统评价和Meta分析无需伦理批准，因为将使用先前发表研究的数据。评价结果将发表在学术期刊上，并在国际会议上展示。 PROSPERO注册号：CRD42024516128。
JMIR Res Protoc. 2024-5-20
Cochrane Database Syst Rev. 2025-5-28
Cochrane Database Syst Rev. 2025-3-24
Trop Med Infect Dis. 2024-9-16
Cochrane Database Syst Rev. 2023-7-20
Healthcare (Basel). 2023-6-16
Ann Lab Med. 2023-9-1
Zotero 插件