Outbreak Detection

The Texas Department of State Health Services (DSHS) Health Service Region 8 (HSR 8) encompasses 28 counties in South Central Texas. Of these, 5 counties are covered by a local health department syndromic surveillance system while the remaining counties fall under HSR 8 syndromic surveillance coverage. Of the 23 counties covered by HSR 8, 15 have hospitals with emergency departments. HSR 8 began receiving emergency department data from 3 hospitals for RedBat® syndromic surveillance monitoring in May of 2006. Four syndromes are monitored daily; Influenza-like Illness, Gastrointestinal Illness (GI), Rash-Illness, and Neurologic-Toxicologic Illness. Aberrations are detected by the Gustav algorithm using RedBat’s ‘Automatic Threshold Alert’ feature. The Gustav algorithm [patent pending], developed by ICPA, Inc., is an advanced variation of the cumulative sum method commonly used for aberration detection. The Gustav algorithm does not require an extended baseline level of illness and is very sensitive to small outbreaks; the algorithm also adjusts for weekly periodicity of medical visits.

Objective

This abstract describes the use of syndromic surveillance at a regional health department to detect an outbreak of norovirus in a nursing home facility.

Objective

To demonstrate how natural language processing (NLP) of clinical records can contribute to case detection and characterization in biosurveillance.

Syndromic surveillance has traditionally been used by public health in disease epidemiology. Partnerships between hospital-based and public health systems can improve efforts to monitor for disease clusters. Greenville Hospital System operates a syndromic surveillance system, which uses EARS-X to monitor chief complaint, lab, and radiological data for the four emergency departments within the hospital system. Combined, the emergency departments have approximately 145,000 visits per year. During March 2007 an increase in invasive group A Streptococcus (GAS) disease in the community lead to the use of syndromic surveillance to determine if there was a concomitant increase in Scarlet Fever within the community.

Objective

 Demonstrate the utility of collaboration between hospital-based and public health syndromic surveillance systems in disease investigation. Demonstrate the ability of syndromic surveillance in identification and evaluation of process improvements.

A U.S. Department of Defense program is underway to assess health surveillance in resource-poor settings and to evaluate the Early Warning Outbreak Reporting System. This program has included several information-gathering trips, including a trip to Lao PDR in September, 2006.

Objective

This modeling effort will provide guidance for policy and planning decisions in developing countries in the event of an acute respiratory illness epidemic, particularly an outbreak with pandemic potential.

This article describes the methodology and results of Team #134ís submission to the 2007 ISDS Technical Contest.

We propose a novel technique for building generative models of real-valued multivariate time series data streams. Such models are of considerable utility as baseline simulators in anomaly detection systems. The proposed algorithm, based on Linear Dynamical Systems (LDS) [1], learns stable parameters efficiently while yielding more accurate results than previously known methods. The resulting model can be used to generate infinitely long sequences of realistic baselines using small samples of training data.

This paper describes a hybrid (event-based and indicator-based) surveillance platform designed to streamline the collaboration between domain experts and machine learning algorithms for detection, prediction and response to health-related events (such as disease outbreaks).

This poster describes the practical integration of Early Event Detection (EED) into the daily operation of a medium sized public health department to improve surveillance for, and response to, outbreaks of communicable disease.

Objective: Emerging and re-emerging infectious diseases (EID/REID) involve large populations at risk and thus they might lead to rapidly increasing cases or case fatality rates. Living in this global village, cross-country or cross-continent spread has occurred more frequently in recent decades, implying that epidemics of any infectious disease can expand from local to national to international if control efforts are not effective.

Current state-of-the-art outbreak detection methods [1-3] combine spatial, temporal, and other covariate information from multiple data streams to detect emerging clusters of disease.  However, these approaches use fixed methods and models for analysis, and cannot improve their performance over time.   Here we consider two methods for overcoming this limitation, learning a prior over outbreak regions and learning outbreak models from user feedback, using the recently proposed multivariate Bayesian scan statistic (MBSS) framework [1]. Given a set of outbreak types {Ok}, set of space-time regions S, and the multivariate dataset D, MBSS computes the posterior probability Pr(H1(S, Ok) | D) of each outbreak type in each region, using Bayes’ Theorem to combine the prior probabilities Pr(H1(S, Ok)) and the data likelihoods Pr(D | H1(S, Ok)). Each outbreak type can have a different prior distribution over regions, as well as a different model for its effects on the multiple streams.  The set of outbreak types, as well as the region priors and outbreak models for each type, can be learned incrementally from labeled data or user feedback.

Objective

We argue that the incorporation of machine learning algorithms is a natural next step in the evolution and improvement of disease surveillance systems. We consider how learning can be incorporated into one recently proposed multivariate detection method, and demonstrate that learning can enable systems to substantially improve detection performance over time.