| Description |
Public health surveillance systems are crucial for the timely detection and response to public health threats. Since the terrorist attacks of September 11, 2001, and the release of anthrax in the following month, there has been a heightened interest in public health surveillance. The years immediately following these attacks were met with increased awareness and funding from the federal government which has significantly strengthened the United States surveillance capabilities; however, despite these improvements, there are substantial challenges faced by today's public health surveillance systems. Problems with the current surveillance systems include: a) lack of leveraging unstructured public health data for surveillance purposes; and b) lack of information integration and the ability to leverage resources, applications or other surveillance efforts due to systems being built on a centralized model. This research addresses these problems by focusing on the development and evaluation of new informatics methods to improve the public health surveillance. To address the problems above, we first identified a current public surveillance workflow which is affected by the problems described and has the opportunity for enhancement through current informatics techniques. The 122 Mortality Surveillance for Pneumonia and Influenza was chosen as the primary use case for this dissertation work. The second step involved demonstrating the feasibility of using unstructured public health data, in this case death certificates. For this we created and evaluated a pipeline iv composed of a detection rule and natural language processor, for the coding of death certificates and the identification of pneumonia and influenza cases. The second problem was addressed by presenting the rationale of creating a federated model by leveraging grid technology concepts and tools for the sharing and epidemiological analyses of public health data. As a case study of this approach, a secured virtual organization was created where users are able to access two grid data services, using death certificates from the Utah Department of Health, and two analytical grid services, MetaMap and R. A scientific workflow was created using the published services to replicate the mortality surveillance workflow. To validate these approaches, and provide proofs-of-concepts, a series of real-world scenarios were conducted. |
