出血性脑卒中患者预后预测及关键因素探索  

Prediction of prognosis and exploration of key factorsin patients with hemorrhagic stroke

作　　者：朱奥宇 朱孟坤 李腾 杨柳 李金红 ZHU Aoyu;ZHU Mengkun;LI Teng;YANG Liu;LI Jinhong(Faculty of Mathematics and Artificial Intelligence,Qilu University of Technology(Shandong Academy of Sciences),Jinan 250353,China)

机构地区：[1]齐鲁工业大学(山东省科学院)数学与人工智能学部,山东济南250353

出　　处：《齐鲁工业大学学报》2025年第2期61-71,共11页Journal of Qilu University of Technology

基　　金：国家自然科学基金(12201333);山东省科技型中小企业创新能力提升工程(2021TSGC1017)。

摘　　要：在出血性脑卒中对人们健康和经济影响重大的今天,早期识别和预后预测十分关键。针对出血性脑卒中发病急、发展快、预后差的问题,主要基于随机森林回归、梯度提升机和多次感知器的知识,通过分析临床患者血肿扩张发生,血肿周围水肿发生及演变规律,构建模型预测出血性脑卒中患者预后情况。在读取数据后进行数据预处理,根据流水号进行数据匹配和数据整合,然后划分数据集,分别使用随机森林回归模型和梯度提升机模型进行训练并预测160名患者90 d的mRS评分;最后使用多层感知器模型进行训练特征信息的关联关系,得到相关数据间的影响效果,为临床相关决策提出建议。实验结果合适有效,未来可应用于金融、城市规划等领域。In today's world where hemorrhagic stroke has a significant impact on people's health and economy,early identification and prognosis prediction are crucial.This article focuses on the urgent onset,rapid development,and poor prognosis of hemorrhagic stroke.Based on the knowledge of random forest regression,gradient boosting machines,and multiple perceptrons,a model is constructed to predict the prognosis of patients with hemorrhagic stroke by analyzing the occurrence and evolution of hematoma expansion and edema around the hematoma in clinical patients.After reading the data,data was preprocessed,data matching and data integration were carried out according to the serial number,and then data sets were divided,and the random forest regression model and gradient elevator model were respectively used for training and to predict the 90-day mRS Scores of 160 patients.Finally,the multi-layer perceptron model is used to carry out the correlation relationship of the training feature information,obtain the influence effect among the relevant data,and put forward suggestions for the relevant clinical decision-making.The experimental results are suitable and effective,and can be applied to finance,urban planning and other fields in the future.

关 键 词：出血性脑卒中 随机森林回归 梯度提升机 多次感知器 特征 

分 类 号：O29[理学—应用数学]