In thinking about health IT, NASA might be the last federal agency that comes to mind. That's what makes a project begun seven years ago that ultimately links the space agency's Jet Propulsion Laboratory (JPL), the National Cancer Institute (NCI) and Children's Hospital Los Angeles a scientific serendipity.

The three organizations each had a common need: to easily share and analyze massive amounts of data from far-flung research centers and laboratories. NASA's aim was planetary science; NCI's was early cancer detection; and Childrens Hospital was building a system to improve clinical decision-support in pediatric practices.

Their odyssey began in 2000 when Dan Crichton from JPL presented a paper on research grid software and planetary science technologies at a National Academy of Sciences conference. While there, he met researchers from NCI who were beginning work on the Early Detection Research Network (EDRN), an effort to take a collaborative approach to the discovery of early detection cancer biomarkers.

The upshot of the meeting was an interagency agreement to explore the use of research grid software developed by JPL to help support the EDRN project.

The JPL grid, called Object Oriented Data Technology (OODT), is a software framework for supporting distributed data search and retrieval. NASA calls the technology "middleware for metadata," a kind of search engine that allows scientists working with data in one expression or format to find and compare their data with another.

"NCI was very interested in the work and hired JPL to consult on health informatics for the EDRN," said Chris Mattmann, a JPL researcher and assistant professor at the University of Southern California, who worked for Crichton. The OODT was a good fit for the project, and eventually became a basic component for both NASA's Planetary Data System and the EDRN.

Meanwhile, back at Childrens Hospital

At about the same time that JPL and NCI were pairing off, Crichton met Dr. Randall Wetzel, director of the Virtual Pediatric ICU at Childrens Hospital.

The two began discussions about the use of the OODT software to support the Virtual Pediatric Intensive Care Unit (VPICU), a next generation infrastructure to support clinical decision support for pediatric critical care.

The VPICU connects emergency rooms, community hospitals and intensive care units worldwide in a virtual network, extending consultations to even the most remote areas. Using OODT, clinicians can access data from a network of pediatric hospitals. The aim of the project: to build an evidence-based foundation for the critical care of children.

"The variability in patients in a pediatric ICU is enormous with regards to age, weight and other factors," said David Kale, a research engineer in the VPICU. "So the question is can we build clinical decision support tools that will help clinicians by augmenting their experience by providing data."

The open source software model has been a key to casting the net for pediatric cancer biomarkers as widely as possible.

"Computer-based tools for medical decision-making have the inherent advantage of capturing the myriad data available in ICUs," Wetzel said, "filtering it for relevance and importance, and providing recommendations based on hundreds of thousands of observations from historical data."

Heather Kincaid, a Northup Grumman consultant and the operations lead for the EDRN Informatics Center at the JPL, is responsible for curating data acquired from studies of cancer biomarkers throughout the United States.

Because of the distributed nature of the research, data curation is a critical aspect of EDRN content management, she said, providing a quality-control gate through which research metadata must pass.

"So many of the same problems exist in other areas," she said. "Since this data is very similar to other types of medical data there are a lot of interesting possibilities."

[This story was updated July 1, 2010, to clarify details on the research partnerships that involved JPL, NCI and Children's Hospital Los Angeles -- the Editor.]