融合网络大数据的重大及新发传染病预警预测模型研究

Deaths of Hepatitis B (HBV) and Tuberculosis (TB) have been ranked as the first two positions of 39 notifiable infectious diseases in China, and emerging infectious diseases such as dengue fever and influenza are too hard to predict and control until outbreak. Aiming at the challenge of infectious disease prevention and control, this project intends to combine data from internet, national surveillance and health records to set up a landscape identification system for characteristics of HBV, TB, dengue fever and influenza by integrating methods and theories from system, big data and landscape epidemiology. We try to estimate incidence and relapse rate of TB, risk probability from HBV to cancer, and early features of emerging infectious diseases. We also establish models which will be used to predict time trend, epidemic and outbreak of infectious diseases while it is used to optimize different measures of prevention and control on the infectious diseases. In comparison with conventional investigation on disease based on only surveillance or survey data, the new investigation supported by integrating big data from landscape sources will put disease into its real world and try to reappear all factors around the disease, which could result in that we can find the risk and make effective strategy. We anticipate that the results from our project will promote new models of prevention and control of infectious diseases driving by big data, and also will contribute to theories of prevention and control on infectious diseases.

乙肝和结核病一直占据我国39种法定传染病死亡率前2位，登革热和流感为代表的新发突发传染病缺少早期预警措施。针对这些传染病防控难题，本项目拟通过整合来自国家监测系统、社区居民健康档案和网络空间的传染病相关大数据，利用系统论、景观流行病和深度学习等多学科研究手段，建立乙肝、结核病、登革热和流感等全景特征辨识网络平台；评估结核病发病率和复发率，评估肝炎演变肝癌的风险概率和干预点，评估登革热和流感爆发早期特征；构建乙肝、结核病、登革热和流感时空动力学传播模型，优选几种传染病的防控策略，实现疾病的早期预警预测。与基于单一物理空间监测数据或调查数据分析的传染病防控模式相比，基于多源大数据分析的传染病防控模式能够将疾病置于其产生的真实背景下进行研究，捕捉更全面、准确、及时的疾病动态演变信息，制定及时有效的针对性防控措施。预期该研究能够促进基于互联网+大数据的传染病防控体系发展，丰富传染病防控理论。

感染者发现不及时造成疫情潜在传播以及新发传染病是我国传染病防控面临的两大挑战问题。课题组在基金支持下，建立了动态乙肝、结核病两种重大传染病和新发突发传染病全景信息辨识网络平台；评估了结核病发病率和复发率，评估了乙肝转变为肝癌的风险概率和干预点；构建了结核病和乙肝的复杂网络时空动力学传播模型，评估了不同防控策略对乙肝和结核病的防控作用；研究了登革热、流感风险特征辨识算法，实现疾病爆发的早期预警预测。在项目资助下，共发表第一标注论文33篇，其中期刊论文18篇，会议论文15篇，有多篇论文发表于Lancet ID等国际顶尖期刊和ICCV等顶级会议；课题共获授权专利2项，在申请专利6项。培养博士9名，硕士30名。项目执行期间，课题负责人及参与人主办研讨会8次，在国内外学术会议上交流学术成果11次。课题负责人受China CDC Weekly杂志邀请作为首位Guest Editor出版传染病模型专刊一期。