Adaboost算法研究土壤微量元素含量与白血病的关系

目的：本文使用数据挖掘方法研究土壤中微量元素和白血病的相关性。方法：使用CFS-Adaboost算法研究我国29个省、市、自治区白血病死亡率的统计数据和土壤中微量元素含量的对应关系。结果：从29种微量元素中发现了13种微量元素与白血病相关,其中砷（As）的相关性较为明显。基于该13种微量元素,建立了土壤中微量元素和白血病致死率的数学模型,该模型的预报准确率可达到89.7%。结论：土壤中微量元素的含量与白血病有一定关系,其中砷（As）元素含量与白血病死亡率较为密切,这与近年文献报导的少量氧化砷治疗白血病效果显著相符合。以上研究发现,可以为研究土壤中微量元素和白血病的关系提供参考,对白血病的防治工作具有一定意义。

Study of Relationship between Trace Elements in Soil and Leukemia Death Rate based on Adaboost Method

Objective： This study was conducted to explore the relationship between trace elements level and leukemia based on data mining methods. Methods： In this article, the relationship between leukemia death rate and trace elements in soil of 29 provinces was studied using the correlation-based feature subset （CfsSubset）-Adaboost method. Results： Thirteen trace elements were found related to leukemia, and Arsenic （As） was found to have impact on leukemia death rate. By using AdaBoost method with the thirteen selected trace elements, the prediction model yields an accuracy rate of 89.7% for the 10-folds cross validation test. It is also found that As has an important impact on leukemia death rate. Conclusion： Trace elements in soil have correlation to leukemia death rate, especially the trace elements Arsenic. Our study results could provide an assistant for investigation of the relationship between leukemia death rate and trace elements, and even could be regarded as a potential supplement for the prevention and clinical treatment of leukemia.