Journal of thoracic disease · 2026
Artificial intelligence in thoracic surgery: a narrative review of clinical advances and applications in 2025.
СтатьяОбзорНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Zhang Yudong, Yang Zheng, Lin Yiluo, Zhao Yulin, Zhou Youtao, Deng Chenyuan, Dai Keyao, Liang Hengrui, Su Yueming
Аннотация
BACKGROUND AND OBJECTIVE: The integration of artificial intelligence (AI) into thoracic surgery accelerated notably over the course of 2025, transitioning from isolated diagnostic aids toward comprehensive clinical pathway integration. The objective of this narrative review is to synthesize the latest evidence on AI applications across the entire thoracic surgical workflow, organized along the patient care continuum from preoperative assessment through intraoperative execution to postoperative management.
METHODS: A PubMed/MEDLINE search was performed on March 6, 2026, and retrieved 378 English-language records published between January 1, 2025 and March 6, 2026. After title and abstract screening, potentially relevant articles underwent full-text review, and studies addressing the clinical applications of AI and related digital technologies across the thoracic surgical pathway were included in this narrative review.
KEY CONTENT AND FINDINGS: In the preoperative domain, large-scale foundation models and computational pathology systems have demonstrated strong performance in nodule risk stratification and noninvasive genomic prediction [area under the curve (AUC) >0.900]. Notably, dual-phase computed tomography (CT) systems such as NeoPred have achieved validation for predicting pathological response to neoadjuvant immunochemotherapy. Intraoperatively, augmented reality (AR) navigation has achieved randomized controlled trial (RCT)-level evidence outperforming conventional localization, while generative AI systems have attained expert-level anatomical recognition for surgical video analytics. Postoperatively, wearable continuous monitoring systems and digital therapeutics (DTx) entered prospective clinical validation. Furthermore, large language models (LLMs) emerged as increasingly important tools for automated surgical documentation. Despite these advances, most studies remain retrospective, and domain shift across institutions limits generalizability.
CONCLUSIONS: While AI has substantially affected thoracic surgery, important gaps persist regarding prospective validation and regulatory governance. Future priorities must focus on prospective multicenter interventional trials linking AI predictions to standardized clinical action protocols, federated learning architectures to overcome data silos, and the development of specialty-specific guidelines building upon the Artificial Intelligence Organization for Next Generation Surgeons (AIONS) 2025 consensus.
METHODS: A PubMed/MEDLINE search was performed on March 6, 2026, and retrieved 378 English-language records published between January 1, 2025 and March 6, 2026. After title and abstract screening, potentially relevant articles underwent full-text review, and studies addressing the clinical applications of AI and related digital technologies across the thoracic surgical pathway were included in this narrative review.
KEY CONTENT AND FINDINGS: In the preoperative domain, large-scale foundation models and computational pathology systems have demonstrated strong performance in nodule risk stratification and noninvasive genomic prediction [area under the curve (AUC) >0.900]. Notably, dual-phase computed tomography (CT) systems such as NeoPred have achieved validation for predicting pathological response to neoadjuvant immunochemotherapy. Intraoperatively, augmented reality (AR) navigation has achieved randomized controlled trial (RCT)-level evidence outperforming conventional localization, while generative AI systems have attained expert-level anatomical recognition for surgical video analytics. Postoperatively, wearable continuous monitoring systems and digital therapeutics (DTx) entered prospective clinical validation. Furthermore, large language models (LLMs) emerged as increasingly important tools for automated surgical documentation. Despite these advances, most studies remain retrospective, and domain shift across institutions limits generalizability.
CONCLUSIONS: While AI has substantially affected thoracic surgery, important gaps persist regarding prospective validation and regulatory governance. Future priorities must focus on prospective multicenter interventional trials linking AI predictions to standardized clinical action protocols, federated learning architectures to overcome data silos, and the development of specialty-specific guidelines building upon the Artificial Intelligence Organization for Next Generation Surgeons (AIONS) 2025 consensus.