[Advances in multi-source medical data-driven identification and spatiotemporal cluster analysis of acute respiratory infections].

Acute respiratory infections (ARIs) are characterized by high morbidity, strong transmissibility, and non-specific clinical manifestations, posing substantial challenges to conventional single-source surveillance systems for early warning. This review systematically summarizes recent advances in ARIs identification and spatiotemporal cluster analysis based on multi-source medical data, focusing on three core technical domains. In terms of medical text information extraction, methodologies have progressively evolved from keyword matching to deep semantic understanding powered by large language models; regarding multimodal medical data fusion, the review covers data-level, feature-level, and decision-level fusion strategies; and in spatiotemporal cluster analysis, both traditional statistical methods and artificial intelligence-based models are discussed. Current research faces key challenges including inconsistent data standards, ambiguous boundaries in data ethics and application scope, incomplete spatial information, and insufficient model interpretability. Future efforts should prioritize advancing medical data standardization and interoperability, and developing hybrid modeling frameworks that balance computational efficiency with interpretability, thereby enabling the transition of ARIs surveillance from passive identification to proactive and precision-oriented early warning.