Grid computing at the Real-time Outbreak and Disease Surveillance Laboratory

by Jeremy U. Espino, M.D.
Director of Information Technology and Open Source Software Development
University of Pittsburgh

Early detection and response to is critical to reducing morbidity and mortality from disease outbreaks. Unfortunately traditional public health surveillance approaches that rely on manual reporting and manual communication methods such as fax and postal mail are not well equipped to provide timely detection of outbreaks. An additional bottleneck is the inability for health departments to readily share disease case data from their database. There is a need for public health software to efficiently gather, exchange, manage, and interpret relevant information from multiple institutions and across state boundaries.

The RODS (Real-time Outbreak and Disease Surveillance) laboratory at the Department of Biomedical Informatics at the University of Pittsburgh has written open-source software that addresses the need for timely outbreak detection. The RODS system is designed to support the collection and management of public health surveillance data including patient visits to hospitals. Real-time data is sent to the RODS system as HL7 messages from Hospitals using secure networks - RODS interfaces directly to the integration engines at hospitals. The system classifies the data, aggregates the data in a relational database system, and applies detection algorithms to look for unusual patterns that may indicate a disease outbreak. If such patterns are detected, the users of the system are alerted. The system provides Web interfaces for users to access the data as well as REST interfaces for programmatic data access.

In a recent CDC-funded effort, the RODS laboratory has been developing modules for the RODS system that facilitate data sharing using the GAARDS and Introduce toolkits and the CDC's GIPSE (Geocoded Interoperable Population Summary Exchange) specification. These new modules will be used to create the Pennsylvania Ohio BIosurveillance Grid (PAOH BIG). This "Grid-enabled" RODS system will enable healthcare departments to securely share notifiable disease cases, vital statistics, syndromic counts with each other. The GIPSE specification, for instance, enables sharing of aggregate information, such as daily counts of respiratory cases in the state in the past month, between departments.

The RODS Laboratory has embraced service-oriented architectures and Grid computing systems to create a federated environment for information sharing. The lab has used the caGrid GAARDS infrastructure for authentication and authorization. They implemented a service proxy to aggregate results from queries using delegated credentials. In this implementation, a client can log onto the environment using Dorian and delegate its credentials to the proxy service via the Credential Delegation Service. The proxy service can interact with the Index Service on behalf of the client to find relevant services and invoke the services and return the results to the client. In order to support finer-grain authorization, the RODS team has implemented query authorization as well.

The team has created a RODS GIPSE service using the Introduce toolkit. This service provides a caGrid-compliant interface to the backend systems based on the RODS infrastructure. The team has also created a RODS GIPSE Client interface using Introduce. This client interface is hosted inside the RODS REST API service. These implementations allow caGrid clients and RODS clients to access services in the other environment, respectively. A caGrid client can access the services and data provided by the RODS GIPSE service. Similarly, a RODS client can access caGrid services, since requests via the RODS REST API are translated to caGrid requests by the RODS GIPSE client.

RODS Laboratory: https://www.rods.pitt.edu/site/

Also see, http://phgrid.blogspot.com/2009/05/paoh-big-notifiable-disease-sharing.html which includes a link to a video clip demonstrating the use of the application.