基于系统流行病学分析策略的疾病因果推断研究

疾病因果机制的探究贯穿生命科学研究之始终且至关重要。新时代大规模人群队列的构建、组学数据获取成本的降低、互联网技术的成熟及高级统计模型的发展，都提示我们未来对于疾病因果关系的探索将日益系统和复杂。本研究将基于系统流行病学的分析策略，以手足口病为例，应用生物信息学研究、流行病学调查、生态学研究等分析方法挖掘、发现疾病相关的影响因素，结合系统综述的研究方法，构建系统性的影响因素框架。在此基础上，基于成组惩罚思想和多核学习方法整合分析多层面数据，识别关键影响因素。接着，进一步应用贝叶斯网络拟合多层面数据间的复杂交互及因果关系，得到基于系统流行病分析策略的因果推断结论，从而实现搭建一套具备实时采集、混合存储、整合建模分析、可视化等功能的大数据处理和分析平台。

Investigation of the causal mechanisms of diseases is essential throughout life course. The construction of large-scale population-based cohorts in the new age, together with the reduction in cost of Omics’ data acquisition, maturity of Internet technology, and the development of advanced statistical models, suggests that the investigation of causal inference will become increasingly systematic and complicated. This study will be based on analytical strategies of systems epidemiology, taking hand, food, and mouth diseases into account. A systematic framework of influencing factors will be built, using the methods of bioinformatics, epidemiological survey and ecological analysis, as well as systematic review. The key risk factors will be selected by integrating and analyzing the multidimensional data based on group punishment and multi-kernel learning methods. Furthermore, a Bayesian network will be applied to fit the complex interaction among selecting parameters to enable robust causal inferences, and obtain conclusions based on the systems epidemiology analysis strategies. Finally, a platform for big data processing and analytics, incorporating real-time acquisition, mixed storage, integration analysis and visualization will be built.