Spatio-temporal Data Mining Based Active Surveillance System Of Infectious Diseases

The outbreak of the infectious disease will directly threat to human’s life and property safety. At present, there are two mainly kinds of surveillance strategies, including active surveillance and passive surveillance. Active surveillance strategy refers to that medical staffs need to search for patients with infectious diseases house by house carefully. Passive surveillance strategy is that people who get infected will go to the doctor. By comparing with the passive surveillance, active surveillance not only can catch data more timely and completely but also can monitor the disease incidences continuously so as to discover the trend of disease outbreak and the distribution of the case in time. However, the research of the theory and application about active surveillance still includes the following problems: 1)How to predict the risk of outbreaks of infectious disease timely and accurately in the case of large monitoring area, strong regional heterogeneity and limited available resources? 2) How to implement a simple, utility and visual active surveillance system of infectious diseases in the rapid development of network technology era, so as to the people can timely know the transmission of the diseases, strengthen the awareness of prevention and avoid the risk of infection?In order to solve the problems above, the main contributions of this paper are as follows:1) We propose a method for real-time active surveillance of malaria with heterogeneous data, including socioeconomic, geographical, surveillance, temperature, rainfall data.Specifically, the dynamic propagation of malaria disease is modeled as a dynamic heterogeneous diffusion network so as to improve the accuracy of the prediction. At the same time, in order to distinguish the regional heterogeneity, we propose a mixture optimization method to predict the risk in a real-time as well as realize the dynamic clustering in different areas, and further to make an accurate prediction of malaria disease.Inthis paper, we explore the ability of the algorithm to deal with incomplete data. By acquiring a part of the cases in a infected area, so as to achieve the prediction of all cases in the infected area.2)We implements the spatio-temporal data mining based active surveillance system of infectious diseases. Based on the method we propose, the system implements the visualization of the active surveillance and real-time prediction. Through the system, people can always view the local outbreaks of the infectious diseases, as well as the future outbreak risk prediction over a period of time. In addition, it contributes to the realization of the infectious disease eliminating from traditional one-way control of government to the universal prevention consciousness in the future. With the cooperation of local hospitals and CDC staffs, we take malaria in Yunnan as a case study, we validate the effectiveness and practicability of the method and the system by using existing real-world data of 2005-2011 in Tengchong county.