文档介绍：基于改进GASVM的智能推荐诊断挂号算法
摘要: 为提高患者就医效率设计了一套智能推荐诊断挂号算法,对大量的历史病案文本进行训练和机器学****以患者特征为依据进行分类并推荐相应的科室。使用遗传算法与支持向量机结合进行特征值提取和参数优化,以核函数参数和文本特征值作为遗传算法的染色体执行选择、交叉和变异操作,为提高遗传算法效率并避免陷入局部最优值,在遗传算法初始化群体阶段使用加权深度优先搜索和***赌结合的机制以保证种群多样性,并对交叉概率和变异概率进行自适应优化,在保留有用遗传信息的同时实现全局搜索。实验结果表明,该算法在有效降低特征值数目的同时提高了分类精度。
关键词: 改进遗传算法; 支持向量机; 智能医疗系统; 智能推荐诊断挂号算法
中图分类号: TN911?34 文献标识码: A 文章编号: 1004?373X(2017)11?0115?04
Intelligent?mendation diagnosis registration algorithm based on improved GA?SVM
CHEN Junmei1, ZHOU Jinyang1, ZHANG Huiying2
(1. Department of Biomedical Engineering, Changzhi Medical College, Changzhi 046000, China;
2. Department of Basic Medicine, Changzhi Medical College, Changzhi 046000, China)
Abstract: An intelligent?mendation diagnosis registration algorithm was designed to improve the efficiency of medical treatment. The intelligent medical system performs training and machine learning for a large number of historical medical record texts, and classifies and mends the appropriate medical departments for patients according to the patient characteristics. The ic algorithm (GA) bined with support vector machine to extract the characteristic value and optimize the parameter. The kernel function parameters and text characteristic values are taken as the chromosomes of the ic algorithm to execute the selection, crossover and mutation operations. To improve the efficiency of GA, and avoid trapping in local optimum, the bining weighted depth?first search with roulette is used in the group initialization stage with GA to guarantee the population diversity
, and performs with adaptive optimization to the crossover probability and mutation probability, which can realize the global search while reserving the useful ic information. The experiment results show that the algorithm can improve the classification accuracy while reducing the quantity of characteristic values.
Keywords: improved ic algorithm; support vector machine; intelligent medical system; intelligent?mendation