Improving disease surveillance

摘要: Traditional disease surveillance systems are instrumental in guiding policymakers’ decisions and understanding dynamics. The first study this dissertation looks at sentinel network design. We consider three locationallocation models: two based on the maximal coverage model (MCM) one K-median model. MCM selects sites that maximize total number of people within a specified distance to site. minimizes sum distances from each individual individual’s nearest Using ground truth dataset consisting million de-identified Medicaid billing records representing eight complete influenza seasons an evaluation function Huff spatial interaction model, we empirically compare networks against existing volunteer-based Iowa Department Public Health influenza-like illness by simulating spread across state Iowa. metrics: outbreak intensity (i.e., burden) timing start, peak, end epidemic). show it is possible design achieves performance identical status quo using fewer sites. also if detection primary interest, actually create matches network’s 42% Finally, effort demonstrate generic usefulness these location-allocation models, examine stroke center selection. describe ineffectiveness current self-initiated approach