代谢组学中机器学习研究进展

以往代谢组学的综述都着眼于对其应用情况进行总结，而对机器学习方法的对比研究不多。为此，在分析讨论目前机器学习方法在代谢组学研究方面最新进展的基础上，对各种机器学习方法及在代谢组学研究上的应用效果进行了总结，给出了代谢组学研究所使用的数据源，比较了各方法的应用效果及其利弊，指出目前代谢组研究中存在的主要问题及可能的改进方案。列举了相关方法在疾病诊断、药物开发等方面的最新应用，展望了代谢组学的研究前景，以期为改进代谢组学研究中机器学习算法的研究工作提供启示和帮助。

Advances of Machine Learning in Metabonomics

The previous review of metabonomics fixes its attention on summarizing the applied situations of meta-bonomies, and yet the research on machine learning approach is not enough. Therefore on the basis of analyzing and discussing the latest advance of the present machine learning approach in metabonomies, the thesis reviews and summarizes many machine learning approaches and their applied effects on the research of metabonomies. After presenting the data source and comparing the applied effects and disadvantages of each approach, it points out the main problems existing in the present metabonomies research and their feasible programmes. Further- more, the thesis enumerates the latest applications of the relative approaches in many aspects, such as disease diagnosis and drug development, and finally presents the outlook of metabonomies. All of these are expected to provide inspiration and aid for improving the research on the machine learning approach in metabonomies.