摘要
为减少因防控不及时导致心力衰竭患者死亡的风险,文中建立一种基于XGBoost的心力衰竭患者死亡风险评价模型,以预测心力衰竭患者未来一年的死亡风险。从患者基本信息和医学检验信息角度分析影响心力衰竭患者死亡风险的因素,应用XGBoost算法建立心力衰竭患者死亡风险预测模型,并使用加拿大多伦多克雷比勒研究所提供的心力衰竭患者临床数据集对模型进行训练和测试。实验结果表明,基于XGBoost的心力衰竭患者死亡风险预测模型在测试集上的MCC为80.78%,ACC为93.33%,AUC为88.20%,F1-score为84.62%。相较于随机森林、支持向量机、梯度提升法这三种算法,文中XGBoost算法在心力衰竭患者死亡风险预测问题上的各项评价指标数据都较为优异,说明所建立的模型有更好的预测能力,可用于实际临床预测。
In order to reduce the death risk of patients with heart failure caused by untimely prevention and control,a death risk assessment model and application of patients with heart failure is established to predict the death risk of patients with heart failure in the next year. The factors affecting the death risk of patients with heart failure are analyzed from the point of view of patients′basic information and medical laboratory information. The death risk prediction model of patients with heart failure is established by means of the XGBoost algorithm,and the clinical data set of patients with heart failure provided by Crabill Institute in Toronto,Canada was used to train and test the model. The experimental results show that the MCC,ACC,AUC and F1-score of the XGBoost-based death risk prediction model for patients with heart failure were 80.78%,93.33%,88.20% and84.62% respectively. In comparison with random forest,support vector machine and gradient lifting algorithm,the XGBoost algorithm has excellent evaluation index data on the prediction of death risk of patients with heart failure,which shows that the built model has better prediction ability and can be used for the practical clinical prediction.
作者
蒋文萍
蒋珍存
董正心
JIANG Wenping;JIANG Zhencun;DONG Zhengxin(School of Electrical and Electronic Engineering,Shanghai Institute of Technology,Shanghai 201418,China;School of Electronic Information and Electrical Engineering,Shanghai Jiao Tong University,Shanghai 201418,China)
出处
《现代电子技术》
2022年第8期155-158,共4页
Modern Electronics Technique
基金
国家自然科学基金项目(61703279)。