人工智能与医疗领域的深度融合，使跨域医学知识整合成为提升智能诊疗水平的关键。然而，不同医学体系间的语义差异与表达隔阂，限制了知识的有效融合与协同应用。文章首先构建了涵盖资源汇聚、知识整合与智能服务的三层框架，以支持从数据融合到知识服务的全流程医疗决策。针对跨域知识语义对齐这一核心难题，文章提出了一种融合通用与领域预训练模型、结合注意力机制与对比学习的深度对齐方法，通过多层级语义提取、动态识别关键信息与对比语义空间优化，实现对异构医学文本间复杂语义关联的精准建模与高效对齐，有效提升了跨域知识对齐的准确性与鲁棒性，为发展能够深度融合多元医学知识的智能决策支持系统提供了技术基础与实践路径。

The deep integration of artificial intelligence and the medical field makes the integration of cross-domain medical knowledge the key to improving the level of intelligent diagnosis and treatment. However, the semantic differences and expression barriers between different medical systems limit the effective integration and collaborative application of knowledge. This paper first constructs a three-layer framework covering resource aggregation, knowledge integration and intelligent services to support the whole process of medical decision-making from data fusion to knowledge services. Aiming at the core problem of cross-domain knowledge semantic alignment, the paper proposes a deep alignment method that integrates general and domain pre-training models, attention mechanisms and contrastive learning, and realizes accurate modeling and efficient alignment of complex semantic associations between heterogeneous medical texts through multi-level semantic extraction, dynamic weight focusing and contrastive semantic space optimization, effectively improves the accuracy and robustness of cross-domain knowledge alignment, and provides a technical foundation and practical path for the development of an intelligent decision support system that can deeply integrate multiple medical knowledge.