Infectious and Tropical Disease Department, Bichat-Claude Bernard Hospital, Assistance Publique-Hôpitaux de Paris, Université Paris Cité, Paris, France; Université Paris Cité and Université Sorbonne Paris Nord, Inserm, IAME, Paris, France.
Medical and Infectious Diseases ICU (MI2) - Bichat-Claude Bernard Hospital, Assistance Publique-Hôpitaux de Paris, Université Paris Cité, Paris, France; EA 7323 - Pharmacology and Therapeutic Evaluation in Children and Pregnant Women, Université Paris Cité, Paris, France.
Infect Dis Now. 2024 Apr;54(3):104864. doi: 10.1016/j.idnow.2024.104864. Epub 2024 Feb 12.
Machine learning (ML) is increasingly being used to predict antimicrobial resistance (AMR). This review aims to provide physicians with an overview of the literature on ML as a means of AMR prediction.
References for this review were identified through searches of MEDLINE/PubMed, EMBASE, Google Scholar, ACM Digital Library, and IEEE Xplore Digital Library up to December 2023.
Thirty-six studies were included in this review. Thirty-two studies (32/36, 89 %) were based on hospital data and four (4/36, 11 %) on outpatient data. The vast majority of them were conducted in high-resource settings (33/36, 92 %). Twenty-four (24/36, 67 %) studies developed systems to predict drug resistance in infected patients, eight (8/36, 22 %) tested the performances of ML-assisted antibiotic prescription, two (2/36, 6 %) assessed ML performances in predicting colonization with carbapenem-resistant bacteria and, finally, two assessed national and international AMR trends. The most common inputs were demographic characteristics (25/36, 70 %), previous antibiotic susceptibility testing (19/36, 53 %) and prior antibiotic exposure (15/36, 42 %). Thirty-three (92 %) studies targeted prediction of Gram-negative bacteria (GNB) resistance as an output (92 %). The studies included showed moderate to high performances, with AUROC ranging from 0.56 to 0.93.
ML can potentially provide valuable assistance in AMR prediction. Although the literature on this topic is growing, future studies are needed to design, implement, and evaluate the use and impact of ML decision support systems.
机器学习（ML）越来越多地被用于预测抗菌药物耐药性（AMR）。本综述旨在为医生提供有关 ML 作为 AMR 预测手段的文献概述。
通过检索 MEDLINE/PubMed、EMBASE、Google Scholar、ACM 数字图书馆和 IEEE Xplore 数字图书馆，截至 2023 年 12 月，确定了本综述的参考文献。
本综述共纳入 36 项研究。32 项研究（32/36，89%）基于医院数据，4 项研究（4/36，11%）基于门诊数据。其中绝大多数研究是在高资源环境中进行的（33/36，92%）。24 项研究（24/36，67%）开发了用于预测感染患者耐药性的系统，8 项研究（8/36，22%）测试了 ML 辅助抗生素处方的性能，2 项研究（2/36，6%）评估了 ML 在预测耐碳青霉烯类细菌定植方面的性能，最后两项研究评估了国家和国际 AMR 趋势。最常见的输入是人口统计学特征（25/36，70%）、先前的抗生素药敏试验（19/36，53%）和先前的抗生素暴露（15/36，42%）。33 项研究（92%）的目标是将革兰氏阴性菌（GNB）耐药性作为输出进行预测（92%）。纳入的研究显示出中等至高度的性能，AUROC 范围为 0.56 至 0.93。
ML 有可能在 AMR 预测方面提供有价值的帮助。尽管关于这个主题的文献正在增加，但仍需要未来的研究来设计、实施和评估 ML 决策支持系统的使用和影响。
