张宁, 玛依拉·阿不都克力木. AI工具在COPD患者继发肺动脉高压早期预测的应用. 2026. biomedRxiv.202607.00013
AI工具在COPD患者继发肺动脉高压早期预测的应用
通讯作者: 玛依拉·阿不都克力木, 1925531959@qq.com
DOI:10.12201/bmr.202607.00013
Application of AI tool in early prediction of secondary pulmonary hypertension in patients with COPD
Corresponding author: mayila·abudukelimu, 1925531959@qq.com
-
摘要:慢性阻塞性肺疾病(chronic obstructive pulmonary disease,COPD)在进展过程中容易合并肺动脉高压(pulmonary hypertension,PH),极大地增加患者出现右心衰竭甚至是死亡的风险,严重影响患者的预后情况。目前临床常用的检查方法尚不能达到对疾病精准预测的目的,临床上迫切需要更优质的新型工具。本文基于人工智能(artificial intelligence,AI)技术在慢性阻塞性肺疾病继发肺动脉高压(chronic obstructive pulmonary disease-pulmonary hypertension,COPD-PH)的早期预测应用方面展开论述,其中包括基于卷积神经网络(convolutional neural network, CNN)的影像学特征的自动量化、多模态数据模型的融合构建,风险层级的划分与预后的理论研究支撑等内容。研究发现AI模型的预测准确率相较于传统方法更高,沙普利性解释(shapley additive explanations,SHAP)也逐步融入模型构建之中。但大部分研究依然存在着缺少外部的前瞻性验证、数据异质性过于显著、多模态融合尚不完善等问题,使得该预测方法在临床实用性上大打折扣。未来的研究中需要继续开展前瞻性多中心验证、联邦学习方法研究、可解释性AI技术的开发等以推动AI工具在COPD-PH早期预测应用中的临床转化。
Abstract: Chronic obstructive pulmonary disease (COPD) is frequently complicated by pulmonary hypertension (PH) during its progression, which substantially increases the risk of right heart failure and mortality, thereby adversely affecting patient prognosis. Currently available clinical examination methods fall short of achieving precise prediction, and there is an urgent need for more effective novel tools in clinical practice. This review discusses the application of artificial intelligence (AI) techniques for early prediction of COPD?PH, encompassing automated quantification of imaging features based on convolutional neural networks (CNNs), construction of multimodal data fusion models, risk stratification, and theoretical underpinnings for prognostic assessment. Studies have demonstrated that AI models achieve higher predictive accuracy than traditional approaches, and Shapley Additive exPlanations (SHAP) are increasingly being integrated into model development. Nevertheless, most studies still face notable limitations, including a lack of external prospective validation, significant data heterogeneity, and suboptimal multimodal fusion, which collectively undermine the clinical utility of these predictive methods. Future research should prioritize prospective multicenter validation, federated learning approaches, and development of explainable AI techniques to facilitate the clinical translation of AI tools for early COPD?PH prediction.
Key words: Chronic Obstructive Pulmonary Disease; Pulmonary Hypertension; Artificial Intelligence; Early Prediction; Radiomics; Deep Learning提交时间:2026-07-07
版权声明:作者本人独立拥有该论文的版权,预印本系统仅拥有论文的永久保存权利。任何人未经允许不得重复使用。 -
图表
-
陈翠妍, 梁会营. 基于多模态数据的慢性阻塞性肺疾病筛查集成模型构建. 2026. doi: 10.12201/bmr.202604.00086
董丹丹, 马红映. 呼出气冷凝液在慢阻肺中的研究价值与进展. 2025. doi: 10.12201/bmr.202512.00038
孙振虎, 周建国, 胡方云. 肺结节CT人工智能诊断的研究进展. 2025. doi: 10.12201/bmr.202504.00068
吴立峥, 王楠, 陈臻. 外周血miR-145-5p、miR-223表达与慢性阻塞性肺疾病合并肺结核患者治疗应答及预后的相关性. 2026. doi: 10.12201/bmr.202604.00163
王文慧, 李琳浩, 王坤, 李鹏, 徐晓晨, 韩萍. 基于“子盗母气”理论探讨慢阻肺合并营养不良的研究. 2026. doi: 10.12201/bmr.202601.00090
.河北大学附属医院消化内科 河北 保定, 杰. 基于深度学习的人工智能技术在胃肠间质瘤中的应用进展. 2024. doi: 10.12201/bmr.202411.00057
王丽萍, 周杰, 肖玮, 孙增涛. 从“三焦气化-肠道菌群代谢”理解肺肠轴在慢性阻塞性肺疾病的作用机制及治疗策略. 2026. doi: 10.12201/bmr.202601.00005
祝可欣, 廖思清, 周霞. 慢性肺曲霉病合并咯血的管理. 2025. doi: 10.12201/bmr.202512.00043
作者简介:△:第一作者:付嘉慧, 杨惠琴. 动脉性肺动脉高压的中医辨证论治研究进展. 2026. doi: 10.12201/bmr.202605.00078
唐宝馨, 高伊丽, 董严芬, 徐川, 王妤杰, 王芳芳, 倪婷, 沈顺怡, 赵春艳. 人工智能赋能早发冠心病患者异常血脂管理的应用现状及策略思考. 2026. doi: 10.12201/bmr.202604.00036
-
序号 提交日期 编号 操作 1 2026-05-18 10.12201/bmr.202607.00013V1
下载 -
-
公开评论 匿名评论 仅发给作者
引用格式
访问统计
- 阅读量:16
- 下载量: 0
- 评论数:0

登录
注册




京公网安备