Data Analytics

Heuristics to detect irregularly shaped spatial clusters were reviewed recently. The spatial scan statistic is a widely used measure of the strength of clusters. However, other measures may also be useful, such as the geometric compactness penalty, the non-connectivity penalty and other measures based on graph topology and weak links.5,6 Those penalties p(z) are often coupled with the spatial scan statistic T(z), employing either the multiplicative formula maximization maxz T(z) ! p(z) or a multiobjective optimization procedure maxz(T(z), p(z)),3,6 or even a combination of both approaches. The geometric penalty of a cluster z is defined as the quotient of the area of z by the area of the circle, with the same perimeter as the convex hull of z, thus penalizing more the less rounded clusters. Now, let V and E be the vertices and edges sets, respectively, of the graph Gz(V, E) associated with the potential cluster z. The non-connectivity penalty y(z) is a function of the number of edges e(z) and the number of nodes n(z) of Gz(V, E), defined as y(z) ¼ e(z)/3[n(z)#2]. The less interconnected tree-shaped clusters are the most penalized. However, none of those two measures includes the effect of the individual populations.

Objective

Irregularly shaped clusters in maps divided into regions are very common in disease surveillance. However, they are difficult to delineate, and usually we notice a loss of power of detection. Several penalty measures for the excessive freedom of shape have been proposed to attack this problem, involving the geometry and graph topology of clusters. We present a novel topological measure that displays better performance in numerical tests.

Public health informatics is an emerging interdisciplinary field that uses information technology and informatics methods to meet public health goals. To achieve these goals, education and training of a new generation of public health informaticians is one of the essential components. AMIA0 s 10 ! 10 program aims to realize the goal of training 10,000 health care professionals in applied health and medical informatics by the year 2010.1 The Department of Biomedical Informatics of the University of Utah was established in 1964. As one of the largest biomedical informatics training programs in the world, the department is internationally recognized as a leader in biomedical informatics research and education.2 The poster hereby describes the collaborative effort between Utah and AMIA to develop a public health informatics online course.

Objective

This poster describes the development and delivery of an online American Medical Informatics Association (AMIA) 10 ! 10 Public Health Informatics course at the University of Utah.

Syndromic surveillance data such as the incidence of influenza-like illness (ILI) is broadly monitored to provide awareness of respiratory disease epidemiology. Diverse algorithms have been employed to find geospatial trends in surveillance data, however, these methods often do not point to a route of transmission. We seek to use correlations between regions in time series data to identify patterns that point to transmission trends and routes. Toward this aim, we employ network analysis to summarize the correlation structure between regions, whereas also providing an interpretation based on infectious disease transmission. Cross-correlation has been used to quantify associations between climate variables and disease transmission. The related method of autocorrelation has been widely used to identify patterns in time series surveillance data. This research seeks to improve interpretation of time series data and shed light on the spatial–temporal transmission of respiratory infections based on cross-correlation of ILI case rates.

Objective

Time series of influenza-like illness (ILI) events are often used to depict case rates in different regions. We explore the suitability of network visualization to highlight geographic patterns in this data on the basis of cross-correlation of the time series data.

Syndromic surveillance data such as the incidence of influenza-like illness (ILI) is broadly monitored to provide awareness of respiratory disease epidemiology. Diverse algorithms have been employed to find geospatial trends in surveillance data, however, these methods often do not point to a route of transmission. We seek to use correlations between regions in time series data to identify patterns that point to transmission trends and routes. Toward this aim, we employ network analysis to summarize the correlation structure between regions, whereas also providing an interpretation based on infectious disease transmission. 

Cross-correlation has been used to quantify associations between climate variables and disease transmission. The related method of autocorrelation has been widely used to identify patterns in time series surveillance data. This research seeks to improve interpretation of time series data and shed light on the spatial–temporal transmission of respiratory infections based on cross-correlation of ILI case rates.

Objective

Time series of ILI events are often used to depict case rates in different regions. We explore the suitability of network visualization to highlight geographic patterns in this data on the basis of cross-correlation of the time series data. 

Electronic disease surveillance systems can be extremely valuable tools; however, a critical step in system implementation is collection of data. Without accurate and complete data, statistical anomalies that are detected hold little meaning. Many people who have established successful surveillance systems acknowledge the initial data collection process to be one of the most challenging aspects of system implementation. These challenges manifest from varying degrees of economical, infrastructural, environmental, cultural, and political factors. Although some factors are not controllable, selecting a suitable collection framework can mitigate many of these obstacles. JHU/APL, with support from the Armed Forces Health Surveillance Center, has developed a suite of tools, Suite for Automated Global bioSurveillance, that is adaptable for a particular deployment’s environment and takes the above factors into account. These subsystems span communication systems such as telephone lines, mobile devices, internet applications, and desktop solutions - each has compelling advantages and disadvantages depending on the environment in which they are deployed. When these subsystems are appropriately configured and implemented, the data are collected with more accuracy and timeliness.

Objective

This paper describes the common challenges of data collection and presents a variety of adaptable frameworks that succeed in overcoming obstacles in applications of public health and electronic disease surveillance systems and/or processes, particularly in resource-limited settings.

Given the periodic nature of influenza activity, it is important to develop visualization tools that enable enhanced decision-making. User-Centered Design is a set of software development methodologies that primarily employ user needs to develop applications. Similarly, Usability Heuristics provide a set of rules that increase the performance of user interfaces, and ease of use. We combined some of these techniques to develop FluView Interactive, a prototype that will enable users to better understand influenza information.

Objective

The objective of this study is to report on the use of User-Centered Design and Usability Heuristics to improve visualization of influenza-related information at the national level. The intention of the prototype is to make data more accessible to different stakeholders including the general public, public health officials at the local and state level, and other experts.

Illnesses like infections, cold, influenza and so on in type 1 diabetes mellitus (T1DM) patients, can compromise the daily patient administered diabetes treatment. This in turn may result in fluctuating blood glucose concentrations, especially hyperglycemia for prolonged periods, which over time can cause serious late complications. The illness prediction project at Tromsø Telemedicine Laboratory aims to construct a prediction model that, through use of patient observable parameters, for example, blood glucose, insulin injections and body temperature, can significantly identify risk of developing illnesses, before onset of symptoms and before illness onset.

Such a model could potentially enable T1DM patients to fight the illnesses, and prepare for an adequate change in the T1DM-management earlier on.

Objective

To develop an illness prediction model that can predict illness in T1DM patients before onset of symptoms, using the patient’s observable parameters.

Dengue is a mosquito-borne viral disease, and there is considerable evidence that case numbers are rising and geographical distribution of the disease is widening within the United States, and around the world. 

The accuracy and reporting frequency of dengue morbidity and mortality information varies geographically, and often is an underestimation of the actual number of dengue infections. As traditional methods of disease surveillance may not accurately capture the true impact of this disease, it is important to gather professional observations and opinions from individuals in the public health, medical, and vector control fields of practice. Prediction markets are one way of supplementing traditional surveillance and quantifying the observations and predictions of professionals in the field. 

Prediction markets have been successfully used to forecast future events, including future influenza activity. For these markets, we divided the possible outcomes for each question into multiple mutually exclusive contracts to forecast dengue-related events. This differed from many previous prediction markets that offered single sets of yes-no questions and used ‘real’ money in the form of educational grants. However, with more detailed contracts, we were able to generate more refined predictions of dengue activity.

Objective

The objective of this project is to use prediction markets to forecast the spread of dengue.

Mandatory notification to public health of priority communicable diseases (CDs) is a cornerstone of disease prevention and control programs. Increasingly, the addresses of CD cases are used for spatial monitoring and cluster detection and public health may direct interventions based on the results of routine spatial surveillance. There has been little assessment of the quality of addresses in surveillance data and the impact of address errors on public health practice.

We launched a pilot study at the Montreal Public Health Department, wherein our objective was to determine the prevalence of address errors in the CD surveillance data. We identified address errors in 25% of all reported cases of communicable diseases from 1995 to 2008. We also demonstrated that address errors could bias routine public health analyses by inappropriately flagging regions as having a high or low disease incidence, with the potential of triggering misguided outbreak investigations or interventions. The final step in our analysis was to determine the impact of address errors on the spatial associations of campylobacter cases in a simulated point source outbreak.

Objective

To examine, via a simulation study, the potential impact of residential address errors on the identification of a point source outbreak of campylobacter.