摘要
目的嗜铬细胞瘤手术切除可能导致术中血流动力学不稳定这一高风险因素发生,会危及生命。本次研究试图通过数据挖掘方法,发现可以预测嗜铬细胞瘤术中发生血流动力学不稳定情况发生的危险因素,为患者术前准备最佳治疗方案提供依据,为临床提供帮助。方法分别使用朴素贝叶斯、决策树(CART、C4.5、C5.0、C5.0 boosted)、随机森林算法、支持向量机(Support Vector Machine,SVM)、Bootstrap-based Aggregate和AdaBoost九种模型,用交叉验证(Crossvalidation)、Hold-out方法和Bootstrap 3种方法,对九种模型分别采取3种划分训练集和测试集的比例,计算准确率。并且计算sensitivity(recall)、specificity、precision、f1-score值进行比较。结果根据结果显示,采用hold out方法划分训练集和测试集,划分比例为6∶4时,AdaBoost模型的准确率最高,为0.8596。而且该模型的specificity、sensitivity(recall)、precision和f1-score指标都是最高的。prevma、ctvalue、bmi、intratime、age、size这6个属性的重要性最大,可以作为预测影响血流动力学不稳定性发生的重要因素。结论数据挖掘技术用于预测嗜铬细胞瘤手术中的IHD风险因素是可行的。将来会越来越多地应用于临床和医学决策,为各种疾病的诊断,治疗和预防提供不断扩展的支持。
Objective Surgical resection of pheochromocytoma may lead to high risk factors for intraoperative hemodynamic instability(IHD),which would be life-threatening.This study aimed to investigate the risk factors that could predict the IHD during pheochromocytoma surgery through the data mining.Methods Nine kinds of data mining algorithm were used in the analysis,including Naive Bayes,decision tree(CART,C4.5,C5.0,and C5.0 boosted),random forest algorithms,Support Vector Machine,bagging and AdaBoost.The cross-validation method,hold-out method,and bootstrap method were used in the validation phase.The accuracy of these algorithms was calculated independently by dividing the training set and the test set.And calculate the sensitivity(recall),specificity,precision,f1-score values for comparison.Results According to the results,the training set and test set were divided by the hold out method,when the ratio reached 6∶4,the accuracy of the AdaBoost model was the highest,0.8596.And the model's specificity,sensitivity(recall),precision and f1-score indicators were the highest.The six attributes of prevma,ctvalue,bmi,intratime,age and size were the most important and could be used as important factors in predicting the occurrence of hemodynamic instability.Conclusion The result of this study showed that data mining may be useful in predicting IHD risk facotrs in the pheochromocytoma surgery.And data mining technologies will be increasingly applied in clinical and medical decision-making,and provide continually expanding support for the diagnosis,treatment,and prevention of various diseases.
作者
赵悦阳
方丽
崔雷
白松
ZHAO Yueyang;FANG Li;CUI Lei;BAI Song(Shengjing Hospital of China Medical University,Shenyang 110004,Liaoning Province,China;Department of Information Management and Information System(Medicine),China Medical University,Shenyang 110001,Liaoning Province,China)
出处
《预防医学情报杂志》
CAS
2020年第2期150-157,共8页
Journal of Preventive Medicine Information
