摘　要：本研究探讨了肠上皮化生（IM）与铜死亡的关系，并构建了一个预测模型。基于生物信息学分析GSE2669数据集中22个IM样本的铜死亡相关基因簇和免疫特征的表达谱，并确定差异表达基因簇。通过评估随机森林（RF）、支持向量机（SVM）、广义线性模型（GLM）及极限梯度提升（XGB）4种机器学习模型的性能，选出了性能最优的RF模型。同时构建了基于4个基因（DMBT1、HPN、SLC13A3 及 YWHAZ）的RF模型，通过表达分析发现，这4个基因与CDX2水平显著相关。实时荧光定量PCR（qRT-PCR）及蛋白质印迹法（Western blot）检测这4个基因在正常胃组织及肠化组织中的表达水平，结果与模型分析高度一致。本研究初步确定了铜死亡在IM中的作用及潜在机制，并成功建立了一个预测模型，为未来的临床应用和研究提供了新的研究方向。
关键词：铜死亡；肠上皮化生；机器学习模型；免疫浸润；预测模型
中图分类号：R735.2                             文献标志码：A                        DOI：10.3969/j.issn.1007-7146.2025.01.008

Abstract: This study examines the association between intestinal metaplasia (IM) and cuproptosis, and develops a predictive model. By analyzing the GSE2669 dataset, gene clusters and immune characteristics associated with cuproptosis in 22 IM samples were characterized, with differential gene clusters identified. The performance of four machine learning models: random forest (RF), support vector machine (SVM), generalized linear model (GLM), and XGBoost (XGB), was evaluated with the RF model showing superior performance. A RF model based on four genes (DMBT1, HPN, SLC13A3, and YWHAZ) was constructed, and their expression was found to be significantly correlated with CDX2 levels. Expression levels of these genes in normal gastric and intestinal metaplastic tissues were confirmed by qRT-PCR and Western blot, aligning with the model’s predictions. This research preliminarily identifies the role and potential mechanisms of cuproptosis in IM and establishes a predictive model, offering new directions for future clinical applications and research.
Key words: cuproptosis; intestinal metaplasia; machine learning models; immune infiltration; prediction model
（Acta Laser Biology Sinica, 2025, 34(1): 063-076）