2026版慢性阻塞性肺疾病全球倡议（Global Initiative for Chronic Obstructive Lung Disease，GOLD）首次将人工智能（artificial intelligence，AI）技术整合进慢性阻塞性肺疾病（chronic obstructive pulmonary disease，COPD）的全程管理框架。AI模型涵盖传统机器学习、深度学习、基础大模型及生成式AI四大类。AI在表型分型、急性加重预警、治疗反应预测等方面展现潜力。通过无监督学习，AI可识别出多种肺气肿影像学分型，不同分型与死亡风险、住院风险等显著相关；深度学习模型在长期和短期急性加重预测中优于传统工具，预警系统可有效降低住院率；机器学习模型在吸入激素有效性预测、肺康复依从性提升及肺减容术成功率预测中表现良好。同时，AI应用面临误诊、算法偏倚、数据隐私及责任归属等伦理挑战，需通过全流程验证、动态知情同意及公平性评估等策略应对。总体而言，AI赋能COPD管理推动从被动治疗向主动预防的范式转变，但仍需在创新、证据与伦理间寻求平衡。

The 2026 Global Initiative for Chronic Obstructive Lung Disease (GOLD) has, for the first time, integrated artificial intelligence (AI) technology into the comprehensive management framework of COPD. AI models encompass four major categories: traditional machine learning, deep learning, large foundation models, and generative AI. They demonstrate potential in phenotyping, early warning of acute exacerbations, and prediction of treatment response. AI identifies multiple imaging subtypes of emphysema through unsupervised learning, with different subtypes significantly associa-ted with mortality risk and hospitalization risk. Deep learning models outperform traditional tools in long-term and short-term prediction of acute exacerbations, and early warning systems can effectively reduce hospitalization rates. Machine learning models perform well in predicting the effectiveness of inhaled corticosteroids, improving adherence to pulmonary rehabilitation, and predicting the success rate of lung volume reduction surgery. Additionally, AI applications face ethical challenges such as misdiagnosis, algorithmic bias, data privacy, and attribution of responsibility, which need to be addressed through strategies including full-process validation, dynamic informed consent, and fairness assessment. Overall, AI-empowered COPD management drives a paradigm shift from passive treatment to active prevention, yet a balance still needs to be achieved among innovation, evidence, and ethics.