How Virginia Tech became the voice of veterinary data
When avian influenza swept through American dairy herds last year, laboratories across the country raced to test milk samples — a specimen type they had never analyzed for the virus before.
The electronic messages reporting those results to the U.S. Department of Agriculture had to travel in hours, not days. They needed codes for new sample types, new testing protocols, and new species combinations. The vocabulary for this crisis did not exist until a small team in Blacksburg built it.
"There were just so many changes because usually the testing is very structured," said Julie Green, who leads the Virginia-Maryland College of Veterinary Medicine's veterinary terminology and standards group. "Avian influenza this last go round just blew that all out of the water. All of a sudden, testing requirements were changing weekly, even daily at times."
When the administration demanded faster data, team members were ready. They'd spent two decades building the infrastructure that enabled rapid reporting.
Building the foundation
"What one might call Disease X is called Disease Y in another system," said Green, director of the veterinary terminology and standards group. The same condition can have dozens of names across different laboratories, regions, and software systems. Standardized terminology eliminates that ambiguity.
The Virginia Tech team is the only organization in the world maintaining this veterinary terminology. Every veterinary organization in the United States and internationally depends on the work, though most have never heard of the team.
Twenty years ago, reporting an outbreak to federal authorities meant printing PDF forms, emailing them to Washington, and waiting while someone at the U.S. Department of Agriculture (USDA) manually entered data into spreadsheets. Months could pass before epidemiologists had complete surveillance pictures. Now, results are transmitted electronically within hours, reaching databases that let officials track disease patterns in near real time.
That transformation required more than faster computers or better internet connections. It required a common language — standardized codes that let a laboratory in California and a database in Maryland communicate.
The veterinary terminology and standards group's origins date back to the early 2000s, when the USDA established the National Animal Health Laboratory Network. Jeff Wilcke and Julie Green, then both at Virginia Tech, recognized that the Systematized Nomenclature of Medicine-Clinical Terms (SNOMED CT) would be essential for electronic messaging between laboratories and federal agencies. But getting new terminology into SNOMED CT took 12 to 18 months at the time — far too slow for responding to disease outbreaks.
So they implemented the newly created extension mechanism, which lets them add codes quickly. When the National Animal Health Laboratory Network needed terminology for a new test or specimen type, the Virginia Tech team could create it in days rather than months.
"We were in the room when the original message was created," Green said. "Jeff Wilcke and colleagues Mike Martin and Jim Case were instrumental in writing that first message structure."
The problem nobody sees
The center maintains something called the veterinary extension of SNOMED CT, or VetSCT. That name means almost nothing to most people, which is part of the problem the team faces when explaining the work.
SNOMED CT is the world's most comprehensive clinical terminology for human medicine — a precisely structured vocabulary used in hospital systems around the globe. When physicians record diagnoses in electronic health records, SNOMED CT provides the standard codes that ensure "heart attack" in Memphis means exactly the same thing as "heart attack" in Miami.
The veterinary extension applies that same principle to animal medicine. When a National Animal Health Laboratory Network facility identifies highly pathogenic avian influenza in a sample, the electronic message that reaches the USDA uses specific LOINC codes for the specimen type, the species tested, the pathogen identified, and the result. Without those codes, systems could not communicate reliably.
Built for speed
In 2014, the International Health Terminology Standards Development Organization, SNOMED's governing body, recognized Virginia Tech's extension as the authoritative global source for all veterinary terminology. Any requests for veterinary terms from anywhere in the world are now handled by the Blacksburg team.
The team created the necessary codes so laboratories could identify and report results during the avian flu outbreak.
The team is small: Green, Wayde Shipman and Katie Krothapalli. They possess an unusual combination of skills — veterinary medical knowledge, programming expertise, and deep informatics training.
"Both of us know computer programming, know SNOMED, and we're both veterinarians, and we've been interested in informatics," Shipman said. "You have very few veterinarians who have that whole variety of skills."
That unique expertise becomes critical as the group's work expands into new areas.
Why it matters now
The group's work is expanding beyond production animal surveillance into areas with even broader implications.
Several veterinary universities — Minnesota, Tufts, California Davis and Colorado State, among them — are attempting to transform their clinical data for research purposes. They are using a common data model that relies on SNOMED terminology to make records from different hospitals comparable and searchable. The goal is to pool veterinary medical data to enable research that no single institution could accomplish on its own.
The challenge is that most veterinary clinical data was not structured for this purpose when it was created. Researchers must map free-text notes and institutional codes to standardized terminology after the fact — a laborious process that the Virginia Tech team has spent years trying to make easier.
"If you want to do research and avoid all these problems, you should structure your data initially," said Shipman, a veterinarian and informaticist on the team.
That principle becomes even more important as artificial intelligence tools enter veterinary medicine. Artificial intelligence (AI) systems work most effectively with clean, structured data. When 47 different names for the same disease occur in a dataset, machine learning algorithms produce less reliable insights.
"Garbage in, garbage out," Green said. "AI is going to make it easier to get the data structured, but you need the standards in place first."
International interest in the Virginia Tech extension has grown to the point where several countries lobbied SNOMED International to provide easier access. Australia, South Korea, and other countries sought to implement VetSCT in their own systems. The result is a migration currently underway that will place the veterinary extension on SNOMED's managed services platform, making it available for download to all member countries.
The vision
The group's current strength lies in production medicine — the USDA contracts that fund most of its work support food-animal surveillance and regulatory messaging. But Green sees a future where standardized terminology transforms companion animal medicine as well.
Only a handful of practice management system vendors have implemented SNOMED terminology in their software. Most veterinarians record notes and diagnoses in free text, making that information effectively invisible to research or artificial intelligence applications.
"I want all of the vendor systems to have at least the option of standardizing codes," Green said. "General practice veterinarians with standard data that they can then decide to contribute to larger pools."
Human medicine achieved this through regulatory pressure — insurance billing requirements and meaningful use of laws forced electronic health record vendors to implement standardized terminology. Veterinary medicine lacks equivalent mandates. Adoption will require demonstrating value to both software vendors and practicing veterinarians.
Green's motivation is clear: "It is our food supply. That is our No. 1 goal. Keeping our food supply safe. Being able to participate in that and help this work more efficiently and quickly and respond quicker — it is important."
