基于局部加权回归和主成分分析的智能预测研究系统  

Intelligent Prediction Research System Based on Locally Weighted Regression and Principal Component Analysis

作　　者：李辉 徐安琪 阳晟君 陈驰 LI Hui;XU Anqi;YANG Shengjun;CHEN Chi(School of Aeronautics and Astronautics,University of Electronic Science and Technology of China,Chengdu 611731,China)

机构地区：[1]电子科技大学航空航天学院,成都611731

出　　处：《移动信息》2025年第1期212-215,共4页Mobile Information

摘　　要：文中通过局部加权回归(LWR)和K-means等机器学习或数学模型方法对脑卒中后血肿周围水肿的变化规律进行智能预测。利用水肿体积数据及检查时间点,构建了分段多项式回归模型和局部加权回归模型。前者通过多个区间的多项式拟合捕捉其体积随时间的非线性变化,后者则更能精细地拟合数据点的局部变化。之后,采用主成分分析(PCA)技术分析了7种不同治疗方法的数据,证明了本文方法在减少水肿体积和控制病情方面具有显著效果。相关性分析显示,水肿体积与血肿体积呈显著正相关关系,不同治疗方法对水肿体积的影响各异。综合这些分析,文中构建了智能预测模型,可有效预测血肿周围水肿的发生和发展。In this paper,ML(Machine Learning)or mathematical model methods such as locally weighted regression(LWR)and K-means are used to intelligently predict the change law of edema around hematoma after stroke.Using edema volume data and inspection time points,a segmented polynomial regression model and a locally weighted regression model are constructed.The former captures the nonlinear change of its volume with time through polynomial matching of multiple intervals,while the latter can better match the local change of data points.After that,principal component analysis(PCA)was used to analyze the data of 7 different treatment methods,demonstrating that the proposed method has significant effects in reducing edema volume and controlling the disease.Correlation analysis showed that edema volume was positively correlated with hematoma volume,and different treatment methods had different effects on edema volume.Based on these analyses,an intelligent prediction model was constructed in this paper,which can effectively predict the occurrence and development of edema around hematoma.

关 键 词：局部加权回归(LWR) 主成分分析 相关性分析 智能预测 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]