基于CT量化评估胸部六个主要解剖单元病征，建立慢性阻塞性肺疾病(chronic obstructive pulmonary disease, COPD)患者5年全因死亡风险预测模型，为开展COPD精准诊治提供参考。

选择2012年12月至2023年9月期间首次于我科住院治疗的733例COPD患者为对象，收集患者的临床资料，基于CT量化评估胸部六个主要解剖单元病征，建立六病征模型(总分0~10分)。采用受试者工作特征(ROC)曲线、K-Means聚类、Kaplan-Meier生存曲线分析。

733例COPD患者中，男575例(78.44%)，女158例(21.56%)，年龄41~93岁，中位年龄为73(65，80)岁。以首次入院日期为基线，278例(37.93%)患者在5年随访节点前死亡。CT量化评估胸部六个主要解剖单元病征结果显示，死亡者与生存者相比，肺动脉直径30.10(26.72, 33.36)比28.00 (24.90, 31.77)mm，改良Reiff评分1(0, 3)比0(0, 2)，Weston评分3(0, 7)比0(0, 3)，竖脊肌密度30.80(20.07, 37.92)比35.22(28.72, 41.94)HU，肺气肿视觉评分(2.91±0.37)比(2.56±0.86)，第4、7、10胸椎平均骨密度121.39(94.65, 150.56)比132.42(105.55, 163.15)HU，胸椎骨折57例(20.50%)比67例(14.73%)，六病征模型得分6(5，7)比5(4，6)，两组差异有统计学意义(P<0.05)。六病征模型预测COPD患者5年全因死亡风险的ROC曲线下面积(AUC值)为0.70；K-Means聚类将患者分为四个评分组；Kaplan-Meier生存曲线提示评分越高，患者5年全因死亡风险越大，高评分组(评分8~10分)患者5年节点时的死亡比例达65%。

通过CT量化评估胸部六个主要解剖单元病征，发现六病征模型可有效预测COPD患者5年全因死亡风险，对早期识别高死亡风险患者具有临床意义。

To establish a 5-year all-cause mortality risk predicting model for patients with chronic obstructive pulmonary disease (COPD) based on quantitative CT assessment of imaging features across six major thoracic anatomic units, thereby informing precision diagnosis and treatment.

A total of 733 COPD patients who were hospitalized in our department for the first time from December 2012 to September 2023 were selected. Clinical data were collected. Chest CT was used to quantify imaging features across six thoracic anatomic units, and the six-feature model (with a total score of 0~10 points) was established. Discrimination was evaluated using the receiver operating characteristic (ROC) curve. Unsupervised K-Means clustering was used to stratify patients by model scores, and Kaplan-Meier survival analysis assessed 5-year all-cause mortality.

Among 733 COPD patients, 575 were male (78.44%) and 158 were female (21.56%), aged from 41 to 93 years, with a median age of 73 (65, 80) years. With the first admission as the baseline, a total of 278 patients (37.93%) died before the 5-year follow-up point. Comparisons of quantified features of thoracic anatomic units revealed statistically significant differences between non-survivors and survivors (P<0.05): pulmonary artery diameter [30.10 (26.72, 33.36) vs. 28.00 (24.90, 31.77)mm], modified Reiff score [1(0, 3) vs. 0(0, 2)], Weston score [3(0, 7) vs. 0(0, 3)], erector spinae muscle density [30.80(20.07, 37.92) vs. 35.22(28.72, 41.94)HU], visual score of emphysema [(2.91±0.37) vs. (2.56±0.86)], average density of the 4th, 7th, and 10th thoracic vertebrae [121.39(94.65, 150.56) vs. 132.42(105.55, 163.15)HU], patient proportion coexisting thoracic vertebral fracture [57(20.50%) vs. 67(14.73%)], and the six-feature model score [6(5, 7) vs. 5(4, 6)]. The area under the ROC curve (AUC value) of the six-feature model for predicting 5-year all-cause mortality risk in COPD patients was 0.70. K-Means clustering analysis separated patients into four score groups, and Kaplan-Meier curves indicated that the higher score, the higher 5-year all-cause mortality risk of COPD patients. The 5-year all-cause mortality of patients at the high score group (with a score of 8 to 10) reached 65%.

Quantitative CT assessment of imaging features across six major thoracic anatomic units enables effective prediction of 5-year all-cause mortality in COPD. The six-feature model can effectively predict the 5-year all-cause mortality risk of COPD patients, which may facilitate early identification of patients at high risk of death, supporting precision management in COPD.