基于常规检验数据的原发性肝癌风险预测模型的建立与评价  

Development and evaluation of risk prediction model for primary liver cancer based on routine laboratory data

作　　者：曹晓强 高颢瑾 杨大干[3] CAO Xinoqiang;GAO Hagjin;YANG Dagan(Department of Laboratory Medicine,the Second ffiliated Hospital of Hainan Med-icine University,Haikou 570311,Hainan;School of Public Healh,Xiamen Uniersity,Xiamen 361104,Frujian;Department of Laboratory Medicine,the First Afiliated Hospial,Schoo of Medicine,Zhejiang Uniersity,Hangzhou 310003,Zhejiang,China)

机构地区：[1]海南医学院第二附属医院检验科,海口570311 [2]厦门大学公共卫生学院,福建厦门361104 [3]浙江大学医学院附属第一医院检验科,杭州310003

出　　处：《临床检验杂志》2023年第8期575-580,共6页Chinese Journal of Clinical Laboratory Science

基　　金：科技创新2030—“新一代人工智能”重大项目(2020AAA0109405)。

摘　　要：目的探索基于常规检验数据的机器学习模型在原发性肝癌风险预测中的价值。方法从医院A收集肝癌组298例和非肝癌组882例,筛选出模型建立的特征参数,建立机器学习的预测模型。从医院B收集肝癌组178例、非肝癌组315例,对所构建的最优模型进行外部验证。结果通过统计学方法筛选出2种最佳特征参数组合,采用机器学习算法分别建立Model1-5和Model6-10,在内部验证集中采用XGBoost算法构建的Model3[ROC曲线下面积(AUC^(ROC))=0.952,准确度=0.899]和Model8(AUC^(ROC)=0.951,准确度=0.897)的性能指标最佳。Model3和Model8共有的特征参数包括性别、年龄、甲胎蛋白、C-反应蛋白、半胱氨酸蛋白酶抑制剂C。Model3的特征参数还有纤维蛋白原,外部验证集的AUC^(ROC)=0.823,准确度=0.793。Model8的特征参数还有清蛋白,外部验证集的AUC^(ROC)和准确度分别为0.816和0.793。结论基于常规检验数据可以构建原发性肝癌的风险预测模型。ObjectiveTo explore the value of the machine learning model based on routine laboratory data for the risk prediction of primary liver cancer.Methods A total of 298 cases of liver cancer group and 882 cases of non-hepatocellular carcinoma group were collected from the hospital A.The characteristic parameters for model were screened out to build a prediction model based on machine learning.A total of 178 cases of the liver cancer group and 315 cases of the non-hepatocellular carcinoma group were collected from the hospital B.The external validation for the constructed optimal model was conducted.Results The two optimal combinations of characteristic parameters were screened out by statistical methods.Model 1-5 and Model 6-10 were built by machine learning algorithm respectively.In the intermal verification set,Model3(AUC^(ROC)=0.952,accuracy=0.899)and Model 8(AUC^(ROC)=0.951,accuracy=0.897)constructed by XGBoost algorithm exhibited the optimal performance indicators.The characteristic parameters shared by Model 3 and Model 8 included sex,age,alpha-fetoprotein,C-reactive protein and cystatin C.The characteristic parameters of Model 3 also included fibrinogen with AUC^(ROC)=0.823 and accuracy=0.793 forthe extenal validation set,and the characteristic parameters of Model 8 also included albumin with AUC^(ROC)=0.816 and accuracy=0.793 for the external validation set,respectively.Conclusion A risk prediction model for primary liver cancer could be constructed based on routine laboratory data.

关 键 词：人工智能 机器学习 原发性肝癌 预测模型 

分 类 号：R446[医药卫生—诊断学]