Data Analytics

Non-temporal Bayesian network outbreak detection methods only look at data from the most recent day. For example, PANDA-CDCA (PC) only looks at data from the last 24 hours to determine how likely an outbreak is occurring. PC is a Bayesian network disease outbreak detection system that models 12 diseases. A system that looks only at each day's data might signal an outbreak one day and not signal it the next. Cooper et al. obtained such results when evaluating the ability of PC to detect a laboratory validated outbreak of influenza. We hypothesized that temporal modeling would attenuate this problem.

Objective

A temporal method for outbreak detection using a Bayesian network is presented and evaluated.

To propose a new spatial scan statistic that has higher ability of pinpointing the true cluster.

This paper describes a methodology for detecting irregular space-time cluster using the space time permutation scan statistic. The methodology includes sequential Monte Carlo simulation and distribution approximation to estimate the error type I.

This paper proposes a new scan statistic which detects disease clusters more accurately than that based on the likelihood ratio.

This paper describes a new expectation-based scan statistic that is robust to outliers (individual anomalies at the store level that are not indicative of outbreaks). We apply this method to prospective monitoring of over-the-counter (OTC) drug sales data, and demonstrate that the robust statistic improves timeliness and specificity of outbreak detection.

The use of spatially-based methods and algorithms in epidemiology and surveillance presents privacy challenges for researchers and public health agencies. We describe a novel method for anonymizing individuals in public health datasets, by transposing their spatial locations through a process informed by the underlying population density. Further, we measure the impact of blurring patient locations on detection of spatial clustering as measured by the SaTScan purely-spatial Bernoulli scanning statistic.

Syndromic surveillance has traditionally been used by public health to supplement mandatory disease reporting. The use of chief complaints as a data source is common for early event detection. Though some public health syndromic surveillance systems allow individual hospitals to view their own data through a web interface, many ICPs have the experience and knowledge-base to conduct their own surveillance and analysis internally. Additionally, they often have interests specific to their hospital which may motivate them to conduct additional syndromic surveillance projects themselves. Lastly, in many cases, ICPs are better able to investigate problems with chief complaint syndrome categorization and aberrations within their own facility before notification of public health staff. A good understanding of the foundation of syndromic surveillance by hospital ICPs can be extremely beneficial when paired with public health to investigate possible cases and outbreaks. ICPs at Greenville Hospital System (GHS), composed of 1110 beds, a level I trauma center with an average of 85,000 visits per year plus three smaller outlying emergency rooms, has had interest in syndromic surveillance for many years and collected data manually for trend analysis using Microsoft Excel to monitor chief complaint data since August 2003.

Objective

Demonstrate the use and benefit to hospital-based infection control practitioners (ICP) of chief complaint data for syndromic surveillance in partnership with public health to assist with traditional public health disease investigations.

We present a new method for multivariate outbreak detection, the ìnonparametric scan statisticî (NPSS). NPSS enables fast and accurate detection of emerging space-time clusters using multiple disparate data streams, including nontraditional data sources where standard parametric model assumptions are incorrect.

To propose a new space-time scan statistic taking overdispersion into account for accurate and timely detection of disease outbreaks.

Our objective in this research is to develop a national, geospatially-explicit set of human agents for use in agent-based models. [The term 'agents', in agent-based modeling, refers to computerized entities that represent individuals who interact with each other and their environment.]