Playing the biosecurity game
Producers unveil their risk behavior in digital field experiments.
As African swine fever marks its territory across parts of Eastern Europe and China, the U.S. swine industry is trying to stay five steps ahead on the chessboard. Implementing more strict biosecurity protocols and eliminating extra visitors and traffic on-farm is crucial to “protect the queen.” However, as researchers at the University of Vermont discovered, it’s really the personnel that are the pawns in the game — and in most circumstances, the weakest biosecurity point.
“The human attitude and response to risk can substantially influence the level of efficacy of biosecurity,” says Gabriela Bucini, a postdoctoral researcher at the University of Vermont. “The aggregate actions by everyone working on and coming onto a farm can make the difference in controlling disease spread, from the single farm all the way to a production system.”
Bucini and her team recently collaborated on a multistate-coordinated agricultural project, funded by the USDA National Institute of Food and Agriculture, that used computer simulations to study the problem from novel perspectives. One of the simulation tools is a swine system model incorporating factors that represent both how disease can spread and how people can behave in response to disease.
“The overarching question is, how does human risk attitude affect the adoption of biosecurity and the ability to control disease spread?” Bucini says.
The researchers decided to examine the behavior of the tactical personnel — the administrators, managers and coordinators on-farm; and the strategic personnel — the president, partners and directors in swine operations. They then worked with students from the University of Vermont Department of Computer Science to create video games called “digital field experiments,” which put players into the business and require them to make decisions.
A controlled means to test human decision-making, game treatments modify information displayed to players to measure how this may influence behavior. Participants are then given economic motivators to increase data accuracy.
“If you play too risky, you may lose everything, so it’s basically what happens in real life,” Bucini says. “You want to have your farm protected, but at the same time there is a cost benefit.”
Do not pass go
Each facility in the game is surrounded by other farms, and each player has a budget. Every month, players must make a decision with the information they have received about biosecurity around their farms and disease challenges in the area. The decision then comes down to how much the player wants to spend out of the budget to protect his or her farm.
“People might play more risk-averse; some people might play riskier because they want to make as much money as possible,” Bucini says. “If you are hit by disease, and you know there was disease in the area, in the game you will lose everything.”
Each player was then given a risk aversion rating, quantified by the monthly decisions from gameplay. Lower scores were given to the players where less biosecurity was adopted, and higher scores were given to players who made a higher number of decisions to increase protection. The scores were computed for each individual over a low and high contagion rate, and clustered using K means, a method of vector quantization.
Risk-averse players (Cluster 1) tended to adopt the most biosecurity earlier in the game. Players who adopted little biosecurity played the riskiest strategy (Cluster 2). Opportunistic players (Cluster 3) adopted little biosecurity at low contagion and increased biosecurity at high contagion.
“The opportunistic group really received the message and used the information about contagion,” Bucini says. “When contagion was low, they played riskier; but as soon as the contagion got higher, they increased biosecurity and played safer.”
How do producers play?
To validate the gameplay decision data, the Vermont researchers recruited 50 swine industry professionals to play the game at the 2018 World Pork Expo in Iowa. Their gameplay was compared to the 50 participants sourced through the Amazon Mechanical Turk crowdsourcing internet marketplace.
According to Bucini and her research colleagues, the producer strategies were very similar to the general population playing. The researchers found a high linear correlation between the ranked player distributions when full visibility of the infection was presented. Although this may not account for all gameplay mechanics and intuition from industry professionals, Bucini says this shows these risk-mitigation decision strategies are realistic for modeling behavior.
“There is high variability in the behavior and strategies in responding to risk. If we want to nudge people towards more safe, risk-averse behaviors and higher biosecurity in the system, we really need to tailor our messages, accounting for the risk attitudes that we see in the population,” Bucini says. “Not all the messages will be received in the same way, so it is important to tailor message to account for the risk attitudes of people.”
Taking data to Level 3
For their next gaming simulation, the Vermont team created a digital swine production system that included a network of movements of hogs and feed, and then they overlaid that production system with an epidemiological model.
For this particular simulation, the researchers chose the porcine epidemic diarrhea virus, as it is known to move through hog and feed networks. They then added a social element to the simulation, assigning a culture or risk attitude to each farm that would be reflected in the biosecurity practices at that farm.
The approach is called agent-based modeling, and the agents in this case are the producers, the slaughter plants, the feed mills and the veterinarians. Producers were given different colors to indicate the kind of hog operation they have, and networks were created among farms, veterinarians, feed mills and slaughterhouses. Then they simulate how PEDV can spread via these networks.
The researchers recognize that the data aren’t perfect, but they do have the mechanics — and it allows them to test “what-if” scenarios.
“What if, in our population of agents, we can shift some of our risk-takers to become more risk-averse? What ultimately happens to the disease spread? What if I am able to control a specific agent to make biosecurity really high; how does the disease spread then?” Bucini asks.
Acting on risk
Bucini says it comes down to identifying the biggest risk-takers.
“If we simulate a producer population dominated by individuals that try to avoid or reduce risk, then biosecurity stays high and can change the course of infection, making disease incidence decrease faster and limiting the number of outbreaks,” Bucini says. “On the other hand, if the population perceives risk to be low, are not consistent with their application of biosecurity practices or simply don’t mind accepting fairly risky situations, then outbreaks are longer and larger, increasing the number of hogs lost.”
Risk communication scholars such as Tim and Deanna Sellnow from the University of Central Florida explain that helping producers internalize the risk they face is essential. “Overall, internalization of the risk and its consequences for both the producers and the larger industry, followed by actionable and clear directions, has the potential to create a more robust and healthy production system,” Bucini says.
About the Author
You May Also Like