How's the Beef?: USDA Goes Mobile

 
 
By Brian P. Watson  |  Posted 2006-04-06
 
 
 

On Christmas Day 2003, officials at the U.S. Department of Agriculture got a gift they wished they could exchange: the positive test results from a cow in Washington State that they believed died from bovine spongiform encephalopathy, better known as mad cow disease.

Until then, technology barely played a role in the agency's efforts to track the disease, a brain-wasting disorder. Inspectors from the department's Animal and Plant Health Inspection Service (APHIS) were testing 1,000 to 2,000 cattle nationwide per month in late 2003 and early 2004, often using pen and paper to record results and observations.

Because USDA officials had not seen mad cow as a major threat here, there were no standard requirements for reporting details from cattle inspections. Once the first case surfaced, however, the agency officials, fearing the possibility that mad cow could spread further, wanted to enable field inspectors to electronically record test results and instantly route them into a database.

In March 2004, three months after that first positive test, the agency directed a software development group based at the Centers for Epidemiology and Animal Health, a division of APHIS' Veterinary Services group, in Fort Collins, Colo., to build a mobile data-collection network to speed up test-taking and aggregating the results. The deadline: 12 weeks. The budget: $1.2 million for equipment and services.

The team met the June 1, 2004 deadline, building a system that uses tablet PCs, form software that allows inspectors to key in or handwrite their inspection reports with a stylus pen, and databases that link inspectors and analysts to both a new mad cow database and the National Animal Health Laboratory Network, an existing database of federal and state records on animal diseases and threats.

"Basically, we had a small amount of time to put together the system, but it's worked well," says Rich Baca, the project's lead developer.

Since 2003, two more cases of mad cow have been discovered in the U.S.—one in Texas in June 2005 and, most recently, in March, a euthanized cow in Alabama. The new system let inspectors access records from the databases about the dead cow's ancestry to see if it was born in another country or if it was born here before certain feed restrictions were put in place; health records and trends documented in the new network led inspectors to quarantine other animals on the farm until they could be tested.

The new mobile computing system supports a testing model created by USDA scientists based on findings from European inspections. There, infections were most common among what scientists dubbed "high-risk" populations of cattle, particularly "downers," or non-ambulatory cows. Using statistical formulas, the USDA scientists believe that if they could test 268,500 cows from the high-risk population of 446,000, they could theorize that less than a handful of cases would exist in the U.S.

Building the System

Since 2001, inspectors with APHIS' Veterinary Services had been inspecting high-risk cows at various surveillance points .

For Baca, the positive test results in 2003—he calls it "the cow that stole Christmas"—coupled with the urgency expressed by USDA officials, meant his team had to introduce new processes for data collection, mobile computing and database management.

"We didn't have a system, per se, because at that point in time [mad cow] had not been identified as needing a major system of surveillance or disease control," explains Dr. Steve Weber, head of the Fort Collins-based Center for Animal Disease Information and Analysis, a division of Veterinary Services that develops and supports technological systems for animal disease detection.

Weber, a principal on the mobile computing project, says that before the 2003 case, inspectors took physical samples at a site and made handwritten notes, sending both to the National Veterinary Services Laboratories in Ames, Iowa.

There, lab workers recorded the results into the facility's information-management system and the National Animal Health Laboratory Network, but with a low priority, given mad cow's then-perceived threat level.

For the new system, with inspectors moving from point to point, Baca and company knew they had to be mobile.

Luckily for them, some of the work was already done. The National Animal Health Laboratory Network was functional, and they gained expertise from previous mobile-computing experiments: an emergency response system in 2003 that used tablet PCs, and another, begun just after the 2003 infection, that had inspectors use laptops running a Lotus Notes-based system that connected to the animal health network.

For the new system, developers chose tablets, because of their greater memory capacity, over handheld devices for inspectors. They deployed 90 tablet PCs from Motion Computing, a contractor that had supplied the agency during the emergency response in 2003.

Weber favors Motion's tablets because of their backlit screen, which is optimal for working in sunny fields; plus, he says, they're rugged enough to perform in unsanitary or unruly sites, like slaughterhouses.

The tablets run software from Mi-Corporation, or Mi-Co, of Research Triangle Park, N.C. Mi-Co customized its software, called Mi-Forms, to allow inspectors to use a stylus to manually note test-site conditions, results and sample characteristics, Baca says.

The Mi-Forms program automatically converts completed forms into eXtensible Markup Language, a text format commonly used to transmit data over the Web. A Mi-Co client program then sends the text files wirelessly to the National Animal Health Laboratory Network, housed in an Oracle 10g database. Field workers now have the ability to wirelessly transmit inspection notes into the database and access health and testing records back from it, which help them decide on what, if any, action to take, like the quarantine on the Alabama farm.

The new capabilities allowed inspectors to speed up their testing regimen—they sampled 176,468 cattle in 2004, more than eight times the 2003 total of 20,543—and hit their goal of 260,000 in January 2005, months ahead of schedule.

USDA Base Case

Headquarters: Veterinary Services, Centers for Epidemiology and Animal Health, 2150 Centre Ave., Fort Collins, CO 80526
Phone: (970) 494-7200
Function: This division of the U.S. Department of Agriculture develops ways to combat health threats to the animal stock in the U.S.
Project Leader: Rich Baca, team leader for Veterinary Services' application information management
Challenge: Build a mobile data-collection system linking inspectors in 50 states to tracking and treatment databases.


BASELINE GOALS:
  • Develop a disease-tracking system and database in 12 weeks.
  • Test more than 260,000 head of cattle over 12 to 18 months, more than 10 times the amount tested in the 2003 fiscal year.
  • Report results to facilities where samples were taken within 24 to 36 hours of testing, effective June 2004.

    : How Now, Down Cow?">
    USDA: How Now, Down Cow?

    The branch of the USDA overseeing mad cow testing and tracking sends inspectors to slaughterhouses, rendering facilities and large farms to gather blood and samples from high-risk cattle, particularly "downers," or those that can't stand. The inspector sends samples to one of seven labs and jots down notes on the animal's description and symptoms, as well as why it's been tested and what samples were taken. Aggregating that information has changed since the first positive test for mad cow. Here's how:

    Pre-Mad Cow Now
    Time frame Until December 2003 June 2004-present
    Reporting tools Pen and paper Tablet PC with Mi-Co software that allows inspectors to hand-write text instead of using keyboard*
    Database entry Manually into a database; lab entered data into its own system Forms software routes to Oracle database holding national records
    Reporting deadlines None;suggested 2 to 3 days 24 to 36 hours

    *Some inspectors still use pen and paper, but now enter notes into a Java application that routes to the same database as the Mi-Co software