文献速递细菌感染、耐药与抗菌药物新发突发、热带与寄生虫病

从床旁问题到复杂诊疗:人工智能在感染性疾病决策中的应用

From bedside queries to complex care: AI for decision-making in infectious diseases.
PubMed 文献 · Clin Microbiol Infect · 2026-08-08

本文综述人工智能在感染性疾病临床决策中的现有及新兴应用,检索PubMed/MEDLINE、Scopus、Web of Science和Google Scholar等数据库。AI应用涵盖大语言模型辅助知识检索、机器学习用于风险分层及治疗工具支持经验性抗生素选择与抗菌药物管理,其临床采用取决于外部验证、工作流整合和临床医生持续监督。

要点

  • 大语言模型可辅助知识检索和常规诊疗问题支持
  • 机器学习模型主要用于风险分层,识别恶化、耐药或复发风险
  • AI治疗工具聚焦经验性抗生素选择、抗菌药物管理和剂量优化

核心结论

  • AI宜视为决策支持层,可扩展专家经验并提高诊疗一致性。
  • 临床采用取决于跨场景外部验证、工作流整合和临床医生持续监督。

研究设计与样本

  • 综述,通过检索PubMed/MEDLINE、Scopus、Web of Science和Google Scholar,并手动筛查相关文献参考文献。

主要发现

  • AI在感染性疾病决策中应用涵盖三大领域:大语言模型辅助知识检索、机器学习风险分层、AI治疗工具支持经验性抗生素选择与抗菌药物管理。
  • 机器学习模型识别临床恶化、抗菌药物耐药或感染复发风险的患者,对分诊和早期干预有潜在意义。

临床怎么用

  • 可将AI作为决策支持层,辅助常规诊疗问题、风险分层和抗菌药物管理。
  • 临床使用需在现有工作流中整合,并保持临床医生监督。

局限性

  • AI在常规诊疗中的作用仍不确定,临床采用依赖跨场景外部验证。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND: Infectious Diseases require rapid decisions based on heterogeneous and evolving clinical data. At the same time, the shortage of infectious diseases specialists continues to widen the gap between demand for expertise and its availability. Artificial intelligence (AI) has emerged as a potential support tool, although its role in routine care remains uncertain. OBJECTIVES: To review current and emerging applications of AI in infectious disease decision-making and to examine their clinical relevance, limitations, and implementation challenges. SOURCES: Recent literature was identified through searches of PubMed/MEDLINE, Scopus, Web of Science and Google Scholar, complemented by manual screening of references from relevant articles on AI applications in infectious diseases. CONTENT: AI applications in infectious diseases span three main areas of clinical decision-making. Large language models can assist with knowledge retrieval and support routine diagnostic and management questions. Machine learning models are primarily used for risk stratification, identifying patients at risk of clinical deterioration, antimicrobial resistance, or infection recurrence, with potential implications for triage and early intervention. AI-supported therapeutic tools focus on empiric antibiotic selection, antimicrobial stewardship, and dose optimization. IMPLICATIONS: AI is best viewed as a decision-support layer that may extend specialist expertise and improve consistency of care. Its clinical adoption, however, depends on external validation across settings, integration within existing workflows, and continued clinician oversight.

来源:Clin Microbiol Infect,查看原文。版权归原作者所有。

引用

  1. Borgonovo F, Pennisi F, Petri F, et al. From bedside queries to complex care: AI for decision-making in infectious diseases. Clin Microbiol Infect. 2026. PMID: 42570802. doi:10.1016/j.cmi.2026.08.002

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。