基于种群密集度的GEP算法挖掘药物不良反应

Mining adverse drug reaction based on GEP of population density

已有的药品安全性评价方法需要知道各参数之间的联系,且过程复杂,或者人为选择统计模型。针对这种情况,提出了药物不良反应(ADR)监测的新方法——基于种群密集度的基因表达式编程算法。该算法在进化中适时计算种群密集度,以判断种群多样性。如果出现搜索迟钝,自动对种群进行新陈代谢,引入新个体;同时算法自动进化ADR人数公式所需的常数,构造ADR人数公式。与标准GEP算法相比,实验表明新方法简单有效,平均进化代数减少约10.8%,平均最高适应度提高约5%。

Existing safety evaluation methods of drugs need know the relation of parameters and manually select statistical model,and the process is complex. View of this,this paper proposed a novel method of ADR monitoring： GEP based on popula-tion density. The algorithm timely computed population density to judge the variety of the populations in evolution. If searching slowly appeared,the populations automatically metabolized and introduced the new individuals. It automatically evolved the constants that the ADR equation needed and constructed the equation. Compared with standard GEP,experiments show that the new method is easy and effective. The average generation is decreased by about 10. 8% and the average top fitness is increased by about 5%.