The current pandemic is far from unprecedented, yet scientists need to be humble when trying to predict what will happen with the COVID-19 virus, Igor Linkov, risk management strategist, told NRT students November 11.
As a Senior Data Analyst working in COVID Response Detail at FEMA and as the Senior Scientific Technology Manager in the U.S. Army Engineer Research and Development Center, Linkov advises government leaders on crises like the COVID-19 pandemic.
“In risk assessment, we assess threat, vulnerability and consequences, and based on that, we develop management strategies,” Linkov said.
In mid-March, the Pacific Ocean Division of the US Army Corps of Engineers called him to Guam and other Pacific islands to assess the threat of COVID-19 and predict its course. The islands, which hold significant U.S. military bases, reported 12 cases of COVID-19. By the end of March, cases appeared limited to one a day, and the governor’s office suggested the crisis might be over and Guam would not need shipments of medical equipment and other resources, Linkov said. He and other scientists ran Susceptible-Exposed-Infectious-Recovered models using data from China and other places. No matter which scenario they used, the SEIR models suggested a major epidemic was coming. They convinced the governor to declare an emergency, and the military started building supply chains to the island.
“If you are in a situation like that with a decision maker asking you to provide information relevant to a decision, you need to frame the problem in ways helpful to them and you need to use your judgment and remember that data will not be black-and-white,” Linkov said. “It’s all shades of gray many times.”
When modelers make interpretations in situations with a big unknown, that can drive significant uncertainty, he said.
“Scientists have limited ability to predict and they need to be humble, especially like in the COVID situation, when we influence life-and-death decisions,” he said.
He first fully realized such limits of scientists, he said, 25 years ago when he did his doctoral research on radionuclides’ migration in forests after the Chernobyl accident. A group of modelers tried to predict radionuclide accumulation in strawberries in a scenario developed by the International Atomic Energy Agency. The modelers’ predictions varied greatly.
“The question is why,” he said. “The answer is not parameter, model or scenario uncertainty. The scenario was quite clear, and the agency was open to answer any questions and share all information on parameters relevant to models. It’s actually how modelers interpret scenarios and make use of the data. This subjective judgment of modelers resulted in huge variations of model predictions.”
This experience showed him that in highly uncertain situations, scientists should use simpler models instead of complex models so that they can parameterize them with limited calibration datasets.
“If you have a simpler model, you have more control and can test hypotheses,” he said. “What else I argue is that, if you are not trying to make absolute prediction but assess courses of action comparatively, simpler models may be better and provide the same degree of reliability.”
In his more recent work on COVID-19, he has been assessing not only risk but also resilience, which he defines as “the ability of systems to recover from threats.”
He is working on resilience issues related to COVID-19 response and recovery and said the supply chain for a vaccine is one example where resilience is key. He has been helping with Operation Warp Speed, the public-private partnership President Trump initiated to produce and distribute COVID-19 vaccines, therapeutics and diagnostics.
“You have heard the news we are about to have a vaccine, but think about producing 600 million doses just for the United States,” he said. “How will you ensure that whatever supply chain develops for that will survive the multiple disruptions likely to occur? So, we argue that you need to assess resilience with the support of network science models.”
He has been analyzing the transportation systems in large U.S. cities and has found that by focusing on efficiency and reducing costs in them, the systems could become less resilient.
“If you reduce costs, you have less slack in the system to be able to recover if components of the system are disrupted,” he said. “So, in the context of transportation, you really need to build alternative roads, and that’s really an expensive proposition.”
The cost of reducing risk may be high, but the United States needs to balance risk and resilience, he said.
“We cannot afford the loss of a significant number of people the way we have experienced now, and we need to be smarter to get out of this pandemic and prevent future pandemics from impacting our society,” he said.
He expressed hope that research on the pandemic will allow scientists to create models that provide a better basis for decisions in the future and noted the precedence of past pandemics.
Among the pandemics he discussed with NRT students was the “Black Death” plague that struck Europe, starting in Venice, Italy, in 1347. Venetians developed what we know as a public health system today by wearing protective clothing such as masks and removing sick people to Lazaretto Island to be quarantined. However, their knowledge of risk was limited, and many believed vampires spread the disease and could chew their way out of graves, so they began to bury victims of the disease with a brick in their mouth.
“Remember this brick in the mouth of vampires,” he said. “We will probably be laughing about what we did in the early days of the pandemic once we know more about COVID. Already, we have reversed the importance of masks from being quite useless, the prevailing notion in the beginning of pandemic, to being the major public health intervention now. So, it just shows how science may not be able to really answer even simple questions under uncertainty.”
He encouraged NRT students interested in working in these areas of risk and resilience to apply for internships with the Army Corps of Engineers. He also said he expects the interdisciplinary education they receive through the NRT to help them in connecting science and policy.
“I think the NRT is teaching you to understand the connections of engineering and policy,” he said. “I tried to present some of the challenges scientists face working in government at the time of crisis. I think the challenges of connecting science and policy will persist and there will always be a disconnect to some extent, but I would really like to encourage you to come work in this domain because I really believe this is what is needed for building a resilient society.”
— Ronica Stromberg, National Research Traineeship Program Coordinator
Collaborating agencies
