吴朝旭 王玥 李潇 南梦蝶 谢美雯 刘常玉 尹亚南 侯丽

摘要 目的：基于80例接受中西醫结合治疗的晚期胃癌患者的临床资料，初步构建晚期胃癌人群类别预测的判别模型。方法：选取2016年12月1日至2018年12月31日来北京中医药大学东直门医院、北京桓兴肿瘤医院、北京三环肿瘤医院就诊的晚期胃癌患者80例作为研究对象，观察随访患者生存时间、一般资料、临床相关资料等，以确诊Ⅳ期后的生存时间为因变量，其他与生存时间相关的变量作为自变量建立Cox模型，筛选具有统计学意义的预后预测因素，并结合前期制定的中医受益人群、非受益人群、中间人群的划分标准，构建晚期胃癌人群类别预测的判别模型。结果：截至2018年12月31日，共纳入80例晚期胃癌患者，死亡62例，脱落18例。构建的线性判别模型对57位患者的类别预测完全正确，总的预测准确率为71.25%。其中，非受益人群的预测准确率最高，为83.33%，其次为受益人群、中间人群，分别为81.08%、59.46%。结论：拟合的线性判别模型对受益人群和非受益人群的识别率较高，对于临床治疗和患者预后的识别具有一定的参考价值。

关键词 胃癌;远处转移;中西医结合;临床资料;受益人群;预测模型;线性判别;生存时间

Establishment of a Discriminant Model for Predicting the Classification of Advanced Gastric Cancer，Based on the Clinical Data of 80 Patients with Advanced Gastric Cancer

WU Chaoxu，WANG Yue，LI Xiao，NAN Mengdie，XIE Meiwen，LIU Changyu，YIN Yanan，HOU Li

（Department of Hematology and Oncology，Dongzhimen Hosptial，Beijing University of Chinese Medicine，Beijing 100700，China）

Abstract Objective：Based on the clinical data of 80 patients with advanced gastric cancer treated with integrated traditional Chinese and Western medicine，this paper attempts to construct a discriminant model for predicting the population type of advanced gastric cancer.Methods：A total of 80 patients with advanced gastric cancer admitted to Dongzhimen Hospital，Beijing University of Chinese Medicine，Beijing Huanxing Cancer Hospital，Beijing Sanhuan Cancer Hospital were selected as the research objects.The patients′ survival time，general data，and clinical related data were observed and followed up.The survival time after diagnosis of stage Ⅳ was used as the dependent variable，and other variables related to survival time were used as independent variables to establish a Cox model to screen for statistically significant prognostic predictors.Combining with the classification standards of TCM beneficiaries，non-beneficiaries，and intermediate populations formulated in the previous stage，a discriminant model for the prediction of the population of advanced gastric cancer was constructed.Results：As of December 31，2018，a total of 80 patients with advanced gastric cancer were adopted，with 62 deaths，and 18 cases of loss.The linear discriminant model was correct for 57 patients，and the overall prediction accuracy was 71.25%.Among them，the prediction accuracy of non-beneficiary group was the highest （83.33%），followed by beneficiary group （81.08%） and middle group （59.46%）.Conclusion：The fitted linear discriminant model has a high recognition rate for the beneficiary and non-beneficiary groups，and has a certain reference value for clinical treatment and patient prognosis.

Keywords Gastric Cancer; Distant metastasis; Integrated Chinese and Western medicine; Clinical data; Beneficiary population; Prediction model; Linear discriminant; Survival time

中图分类号：R242文献标识码：Adoi：10.3969/j.issn.1673-7202.2021.03.007

胃癌是我国最常见的恶性肿瘤之一。根据国家癌症中心2019年公布的数据，胃癌高居我国男性恶性肿瘤发病率的第2位、女性恶性肿瘤的第5位;死亡率分别居第3位、第2位[1]。由于起病隐匿以及大规模普查的缺失，我国早期胃癌的检出率不足10%[2]，这预示大多数患者在确诊时已失去了最佳的手术时机。……