基于加权混合核函数的KPCA-PLS中药量效数据回归分析
(江西中医药大学,江西 南昌 330004)
摘要
中药剂量-效应数据往往具有样本量小、变量间相关性强、非线性关系复杂等特点,传统多元回归方法通常难以满足该类数据的分析需求。针对上述问题,本文融合核主成分分析(KPCA)与偏最小二乘(PLS)各自的优势,提出一种基于特征加权的多核KPCA-PLS回归方法。该方法首先利用偏最小二乘变量投影重要性(PLS-VIP)度量特征重要性并计算权重,将其引入后续的核函数构造中,以增强模型对关键变量的敏感性;其次,在KPCA中构建融合RBF核与多项式(Poly)核的多核学习结构以提取主成分,从而兼顾局部与全局特征提取能力;随后将KPCA提取的主成分输入PLS模型进行回归,并引入多种群遗传算法,在迭代过程中自动优化核函数参数组合,以获得最优回归效果。在两个中药数据集和3个公开数据集上,与6种其他方法进行对比实验,结果验证了所提方法的优越性。
关键词
中药量效数据;特征加权;核主成分;偏最小二乘
参考文献
- [1] Zhao L,Zhang H,Li N,et al. Network Pharmacology, A Promising Approach to Reveal the Pharmacology Mechanism of Chinese Medicine Formula [J/OL].Journal of ethnopharmacology,2023,309:116306.https://doi.org/10.1016/j.jep.2023.116306.
- [2] Korchevskiy A A,Korchevskiy A. Non-linearity in Cancer Dose-Response: the Role of Exposure Duration [J/OL].Computational Toxicology,2022,22:100217.https://doi.org/10.1016/j.comtox.2022.100217.
- [3] He J X,Huang L W,Xiao Y,et al. Prediction model of Continuous Discharge Coefficient from Tank Based on KPCA-DE-SVR [J/OL].Journal of Loss Prevention in the Process Industries,2024,89:105316.https://doi.org/10.1016/j.jlp.2024.105316.
- [4] Li X H,Jia R C,Zhang R R,et al. A KPCA-BRANN Based Data-Driven Approach to Model Corrosion Degradation of Subsea oil Pipelines [J/OL].Reliability Engineering & System Safety,2022,219:108231.https://www.sciencedirect.com/science/article/abs/pii/S0951832021007092?via%3Dihub.
- [5] Schölkopf B,Smola A,Müller K R. Nonlinear Component Analysis As A Kernel Eigenvalue Problem [J].Neural Computation,1998,10(5):1299-1319.
- [6] Wold S,Sjöström M,Eriksson L. PLS-regression: A Basic tool of Chemometrics [J].Chemometrics and Intelligent Laboratory Systems,2001,58(2):109-130.
- [7] Liu X Z,Yang X,Guo R X. Regional Differences in Fossil Energy-Related Carbon Emissions in China’s Eight Economic Regions: Based on the Theil Index and PLS-VIP Method [J/OL].Sustainability,2020,12(7):2576.https://doi.org/10.3390/su12072576.
- [8] Cheng K,Lu Z Z,Wei Y H,et al. Mixed kernel Function Support Vector Regression for Global Sensitivity Analysis [J].Mechanical Systems and Signal Processing,2017,96:201-214.
- [9] 陈渤,刘宏伟,保铮.一种融合核优化算法 [J].西安电子科技大学学报,2007(4):509-513,582.
- [10] Forrest S. Genetic algorithms [J].ACM computing surveys (CSUR),1996,28(1):77-80.
- [11] Liu K L,Lin T,Zhong T T,et al. New Methods Based on A Genetic Algorithm Back Propagation (GABP) Neural Network and General Regression Neural Network (GRNN) for Predicting the Occurrence of Trihalomethanes in tap Water [J/OL]. Science of The Total Environment,2023,870:161976.https://doi.org/10.1016/j.scitotenv.2023.161976.
- [12] Wei Z Q,Lu T,Gu R P,et al. DBN-GABP model for Estimation of Aircraft Wake Vortex Parameters Using Lidar Data [J].Chinese Journal of Aeronautics,2024,37(9):347-368.
- [13] Li Y,Jia M,Han X,et al. Towards a Comprehensive Optimization of Engine Efficiency and Emissions by Coupling Artificial Neural Network(ANN)with Genetic Algorithm (GA) [J/OL].Energy,2021,225:120331.https://doi.org/10.1016/j.energy.2021.120331.
- [14] Ranstam J,Cook J A. LASSO Regression [J].Journal of British Surgery,2018,105(10):1348-1348.
- [15] Smith P F,Ganesh S,Liu P. A Comparison of Random Forest Regression and Multiple Linear Regression for Prediction in Neuroscience [J].Journal of Neuroscience Methods,2013,220(1):85-91.
- [16] Li J,An X,Li Q,et al. Application of XGBoost Algorithm in the Optimization of Pollutant Concentration [J/OL].Atmospheric Research,2022,276:106238.https://www.sciencedirect.com/science/article/abs/pii/S0169809522002241?via%3Dihub.
- [17]曾红,周秋贵,罗婷,等.厚朴酚与和厚朴酚对小鼠腹泻及胃肠排空抑制的影响比较[J].中药材,2015,38(10):2160-2162.
- [18]彭昕欣,易鑫,刘婧,等.肠道内容物和黏膜菌群对枳实导滞汤剂治疗食积便秘的响应差异[J].中国微生态学杂志,2023,35(11):1241-1247,1256.
基金项目
国家自然科学基金(82260849);江西中医药大学校级科技创新团队发展计划(CXTD22015)
作者简介
李想(2001—),男,汉族,安徽铜陵人,硕士研究生,研究方向:医药数据挖掘;通信作者:聂斌(1972—),男,汉族,江西峡江人,教授,硕士生导师,博士研究生,研究方向:医药数据挖掘、中医药信息学、中药学;郑水飞(1998—),男,汉族,江西上饶人,硕士研究生,研究方向:医药数据挖掘;章祉瑶(2001—),女,汉族,江西南昌人,硕士研究生,研究方向:医药数据挖掘;孟冲(2003—),男,汉族,安徽阜阳人,硕士研究生,研究方向:医药数据挖掘;史晨浩(2001—),男,汉族,江苏南通人,硕士研究生,研究方向:医药数据挖掘。