预测肺腺鳞癌患者骨转移机器学习模型的建立  

Establishment of machine learning model for predicting bone metastasis in patients with lung adenosquamous carcinoma

作　　者：朱英浩 王诗淇 张纬 刘瑜[1,2] ZHU Yinghao;WANG Shiqi;ZHANG Wei;LIU Yu(Department of Cardiothoracic Surgery,the First Affiliated Hospital of Wenzhou Medical University,Wenzhou 325035,China;The First School of Medicine,School of Information and Engineering,Wenzhou Medical University,Wenzhou 325035,China;Department of Internal Medicine,Niansanli Park of the Forth Affiliated Hospital of Zhejiang University,Jinhua 322000,China)

机构地区：[1]温州医科大学附属第一医院心胸外科,浙江温州325035 [2]温州医科大学第一临床医学院(信息与工程学院),浙江温州325035 [3]浙江大学医学院附属第四医院廿三里院区内科,浙江金华322000

出　　处：《温州医科大学学报》2023年第7期588-594,F0003,共8页Journal of Wenzhou Medical University

基　　金：浙江省自然科学基金项目(LY21H160059)。

摘　　要：目的:探究机器学习算法在构建肺腺鳞癌患者骨转移预测模型中的价值。方法:选取监测,流行病学和最终结果(SEER)数据库以及温州医科大学附属第一医院2017年1月1日至2021年12月31日术后确诊为肺腺鳞癌的患者数据。为了建立预测模型,应用了随机森林(RF)、支持向量机(SVM)、极端梯度提升(XGBoost)、梯度提升(GBM)、神经网络(MLP)和K近邻(kNN)6种算法。采用受试者工作特征(ROC)曲线来评价模型的预测能力。结果:获取SEER数据库中1 919例及温州医科大学附属第一医院51例符合条件的患者的医疗记录。机器学习模型的结果显示,肺腺鳞癌转移到骨以外的器官及其淋巴结转移是骨转移的最重要预测因素(P<0.01)。基于XGBoost算法的机器学习模型的属性更加准确和高效。结论:肺腺鳞癌的骨转移会导致预后不良。基于机器学习的预测模型能够早期准确地预测肺腺鳞癌患者发生骨转移的可能性,对临床决策有重要意义。Objective:To explore the value of machine learning algorithm in the prediction model of lung adeno-squamous cell carcinoma bone metastasis.Methods:Data of patients with lung adeno-squamous cell carcinoma were obtained from the Surveillance,Epidemiology and End Results Database(SEER)and the First Affiliated Hospital of Wenzhou Medical University between January 2017 and December 2021.In order to build the prediction model,six algorithms including random forest(RF),support vector machine(SVM),eXtreme Gradient Boosting(XGBoost),Gradient Boosting Machine(GBM),Multi-Layer Perceptron(MLP)and KNearest Neighbor(kNN)were applied.ROC curve was used to evaluate the prediction capability of these models.Results:Medical records of 1,919 eligible patients were obtained from the SEER database,and 51 obtained from the First Affiliated Hospital of Wenzhou Medical University.Machine learning model results showed that lung adenosquamous metastasis to organs other than bone and lymph node metastasis were the most important predictors of bone metastasis.The attributes of machine learning model based on XGBoost algorithm were more accurate and efficient.Conclusion:Bone metastasis of lung adenosquamous cell carcinoma leads to poor prognosis.The predictive model based on machine learning can accurately predict the possibility of bone metastasis in patients with lung adenosquamous cell at early stage,which is of great significance in clinical decision-making.

关 键 词：肺腺鳞癌 骨转移 机器学习 预测模型 

分 类 号：R654.2[医药卫生—外科学]