基于tlPSO-SVM模型的肿瘤进展预测  

Tumor Progression Prediction Based on tlPSO-SVM Model

作　　者：方丽英[1] 陈培煜 王普[1] 李爽[1] 杨建栋[1] 万敏[1] 

机构地区：[1]北京工业大学电子信息与控制工程学院,北京100124

出　　处：《计算机工程》2014年第7期189-192,共4页Computer Engineering

基　　金：北京市委组织部优秀人才培养计划基金资助项目(2010D005015000001);北京市新世纪百千万人才工程基金资助项目;北京市教委科研计划基金资助面上项目(KM201410005004)

摘　　要：为研究患者肿瘤进展情况与各项指标之间的关系,以支持向量机(SVM)作为分类模型,根据各项检查指标预测肿瘤进展情况。设计三层粒子群优化算法(tlPSO)对SVM模型进行参数寻优,使用训练集建立分类模型,利用测试集评估模型性能,得到tlPSO-SVM模型。tlPSO算法能有效降低陷入局部最优解的机率,获取全局最优参数,从而使模型具有最优的分类性能。将血常规、中医症候、FACT评分等指标作为输入,肿瘤进展情况作为分类输出,建立分类模型并进行预测。实验结果表明,tlPSO-SVM模型准确率较高,具有较好的分类性能。For researching the relation between the lung cancers’ tumor progression and the medical factors, taking Support Vector Machine（SVM） as the classification model. Firstly, utilizing the improved Particle Swarm Optimization（PSO） algorithm-three layers Particle Swarm Optimization（tlPSO）, to optimize the SVM parameters; Secondly, establishing the classification model; Finally, using the test set to evaluate the model. The tlPSO algorithm reduces the probability of falling into local optimal solution by the algorithm, obtained the global optimal conclusion and makes the model with optimal performance. Choosing blood routine, the TCM symptoms and the FACT value as the input of the experiment, the tumor progression as the classification output, the classification model is established for prognosis. From the experiment, the improved PSO is superior in finding the optimal parameters and improved the accuracy of classification model, and the proposed tlPSO-SVM model has better classification performance.

关 键 词：肿瘤进展 粒子群优化算法 支持向量机 参数寻优 分类模型 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]