ISDS Conference

In a classical surveillance system one looks for disturbances in the number of cases, but in a spatio-temporal system, not only the number of cases observed but also where they are located is reported. What location is reported, and to which degree of accuracy it is reported are important. At one extreme les near-perfect information about each case, as with contact tracing; at the other extreme we have no information about location; viz. just that the patient exisits, or a temporal system. From maximum spatial precision to no spatial precision, one gains in speed of reporting and privacy; but one loses power to detect outbreaks. For example, in Ozonoff et al. we see that more than one address is better than just a single one. This general point is intuitively appealing, and can be demonstrated. 

Objective

This paper quantifies the effect of not providing full information about the location of patients when dealing with spatio-temporal systems in syndromic surveillance. The study investigates the loss of power to detect clusters when aggregation takes place. 

BioSense is a national system that receives, analyzes, and visualizes electronic health data and makes it available for public health use. In December 2007 CDC added the Influenza Module to the main BioSense application.

Objective

This presentation describes the new BioSense Influenza Module, its performance during the 2007-8 influenza season, and modifications for the 2008-9 influenza season.

One of the emerging priorities for the use of syndromic surveillance is for the monitoring of environmental health conditions. Heat-related illness (HRI) is of growing public health importance, particularly with climate change and anticipated increased frequency of heat waves. High ambient temperatures are responsible for significant morbidity and mortality, as was demonstrated during the 2003 heat waves in Europe that resulted in an estimated 45,000 excess deaths. A syndromic surveillance system that is able to detect early indications of excess HRI may start the public health response earlier, and thus reduce associated morbidity and mortality. Our research group is exploring the potential use of 911 medical dispatch data for the surveillance of HRI in Toronto. An important step in this assessment is exploring the association between temperature and 911 dispatch calls for HRI.

Objective

This paper describes the association between 911 medical dispatch calls for heat-related illness and maximum temperature in Toronto, Ontario during the summer of 2005.

The New York State Department of Health (NYSDOH) currently applies EARS’s CuSum analyses to Medicaid Over the Counter and Prescription Medications data obtained from the Office of Medicaid Management's data warehouse. Daily drug category counts are compared with counts for a 7-day baseline period to generate C1, C2, and C3 signals for 62 counties and 8 Syndromic Surveillance Regions. Summary tables and graphs are posted on the NYSDOH Secure Health Commerce Network for access by state, regional, and county users.

The 7-day baseline CuSum method of analysis of syndromic surveillance databases can result in the generation of a large number of signals. Many signals are generated for counts that, upon manual review of 30-day or long-term trend graphs, are clearly within the range of normal daily variation and would not require follow up by public health authorities.

In order to prevent user desensitization to generated signals and minimize NYSDOH Syndromic Surveillance System end-user burden, supplemental measures that would indicate a daily count is higher than expected are currently being investigated.

Objective

To supplement CuSum analyses of syndromic surveillance databases within NYSDOH's Electronic Syndromic Surveillance System with other measures that would indicate a daily count is higher than expected in order to minimize the end-user burden of following up generated signals.

Objective

There were two objectives of this analysis. First, apply text-processing methods to free-text clinician notes extracted from the VA electronic medical record for automated detection of Influenza-Like-Illness. Secondly, determine if use of data from free-text clinical documents can be used to enhance the predictive ability of case detection models based on coded data.

Graph theory concepts are well established in epidemiology, with particular success as a description of agent-based modeling. An agent-based viewpoint leads to conclusions about the spatial distribution of links: infection is more likely among individuals in close proximity. In this analysis, we seek evidence of these temporal-spatial links though the properties of random geometric graphs.

Our investigation begins with the interpoint distance distribution (IDD) approaches referenced, which provide a promising approach to detect outbreaks that are localized in both space and time. Using a Mahalanobis-based metric, this distribution is compared to an expected distribution derived from historical records.

Unfortunately, when applied to a complex data set such as from Children’s Hospital Boston, the IDD provides inadequate power. Emergency Department chief complaints from 1/1/2000-12/31/2004 were used to identify patients with infectious respiratory illness based on a triage process.

As in most realistic catchments, the historic density of patients varies greatly over the catchment area.

Objective

This paper uses geometric random graph concepts to develop early detection algorithms for the real-time detection and localization of outbreaks.

National surveillance is used to detect the emergence and spread of influenza virus variants and to monitor influenza-related morbidity and mortality. Nurse telephone triage (“call”) data may serve as a useful complement to traditional influenza surveillance, especially at times or in places traditional surveillance is not operating. It may also be useful to detect increased occurrence of non-influenza respiratory infection.

Objective

We compared state-level nurse call data to CDC national influenza surveillance data to determine how well call data performed relative to CDC sentinel provider and viral isolate data. This quantitative analysis extends an earlier semiquantitative regional analysis of the same data.

Previously we used an “N-Gram” classifier for syndromic surveillance of emergency department (ED) chief complaints (CC) in English for bioterrorism. The classifier is trained on a set of ED visits for which both the ICD diagnosis code and CC are available by measuring the associations of text fragments within the CC (e.g. 3 characters for a “3-gram”) with a syndromic group of ICD codes. Because the ICD system is language independent, the technique has the potential advantage of rapid automated deployment in multiple languages. Our objective was to apply the N-Gram method to a training set of Turkish ED data to create a Turkish CC classifier for the respiratory syndrome (RESP) and determine its performance in a test set.

Objective

To determine how closely the performance of an ngram CC classifier for the RESP syndrome matched the performance of the ICD9 classifier.

T-Cube is especially useful for rapidly retrieving responses to ad-hoc queries against large datasets of additive time series labeled using a set of categorical attributes. It can be used as a general tool to support any task requiring access to such data. From the application’s perspective it is transparent: it acts just like the database itself, but an incredibly quickly responding one. The authors had a chance to put T-Cubes into practical use as an enabling technology in applications requiring massive screening of multidimensional temporal data. These applications include two systems to support monitoring of food and agriculture safety and predictive analytics developed at the US Department of Agriculture and the Food and Drug Administration, as well as a system to monitor and forecast health of a fleet of aircraft operated by the US Air Force.

Objective

T-Cube, a data structure designed to efficiently represent large collections of temporal data has been shown to benefit surveillance applications involving monitoring sales of over-the-counter medications and emergency department visits. In this paper we present efficiencies which can be realized in practical applications of T-Cube beyond its original areas of deployment, and we advocate a widespread use of it as a technology which makes manual ad-hoc lookups as well as many kinds of complex automated analyses feasible.

Varied approaches have been used by syndromic surveillance systems for aberration detection. However, the performance of these methods has been evaluated only across a small range of epidemic characteristics.

Objective

We conducted a large simulation study to evaluate the detection properties of 6 different algorithms across a range of outbreak characteristics.