Reimagining our pandemic problems with the mindset of an engineer

Every dog has become an amateur statistician and epidemiologist over the past 20 months. A group of bona fide statisticians and epidemiologists came to the conclusion that pandemic issues could be solved more efficiently if they adopted an engineer mindset. This means that they focused on practical problem-solving using an iterative, adaptive strategy.
The researchers recently wrote an essay entitled Accounting for uncertainty during pandemics. They reflect on the roles they play during a crisis and suggest ways to be more prepared for the next one. They suggest that the answer lies in reimagining epidemiology from an engineering and less science perspective.

Public health policy is informed by epidemiological research. It also has an inherent mandate to protect and prevent. The pandemic proved to be a troubling example of how difficult it was to find the right balance between pragmatic solutions and pure research findings.

Practical decisions are what we have to make, so how important is uncertainty? Seth Guikema

Jon Zelner, coauthor of this essay, said that I had always imagined that epidemiologists would be helpful in times of crisis. Our role was more complicated and poorly defined than I expected. Zelner, a coauthor of the essay, said that he witnessed a flurry of research papers. Many of them had little to no impact on the actual situation.

Zelner said that there were many missed opportunities due to missing links between the ideas, tools and worlds epidemiologists suggested and were meant help.

Give up on certainty

Andrew Gelman (a Columbia University statistician and political science scientist) provided the larger picture in the essay introduction. He compared the pandemics outbreak by amateur epidemiologists to war making every citizen a war strategist and geographer. Instead of maps with colored pins and charts of exposure and deaths, people are arguing about herd immunity and infection fatality rates the same way they debated wartime alliances and strategies in the past.

Are masks still needed despite all the data and public discourse? The uncertainty poured in.

The researchers, which also included Ruth Etzioni from the University of Washington as well as Julien Riou from the University of Bern, attempted to reenact the events. The tools that were used to address problems such as the estimation of the transmission rate from person to person or the number of cases per population at any one time were examined. They evaluated everything, from data collection (the quality and interpretation of the data were the most difficult challenges of the pandemic), to model design, statistical analysis, communication, trust, and decision-making. They wrote that uncertainty is present at every step.

Gelman states that the analysis doesn't capture enough of my confusion during those first months.

Statistics is one way to combat uncertainty. Gelman sees statistics as mathematical engineeringmethods that allow for measurement and discovery. Statistics aims to shed light on the world with an emphasis on uncertainty and variation. Iterative processes should be initiated when new evidence is available. This will allow for refinement of existing knowledge and increase certainty.

Science is humble and can be refined in the face uncertainty. Marc Lipsitch

Susan Holmes, a Stanford statistician who wasn't involved in the research, sees parallels to the engineering mindset. Engineers are constantly updating their picture. She says, revising as new data and tools become accessible. An engineer will offer a first-order approximation, which is more general and less focused, in order to solve a problem.

Gelman, however, has previously warned that statistical science can be deployed as a machine for laundering uncertaintydeliberately or not, crappy (uncertain) data are rolled together and made to seem convincing (certain). Statistics used to combat uncertainties are often misrepresented as a kind of alchemy that can transform them into certainty.

This was what we saw during the pandemic. In an era of chaos and uncertainty, statisticians and epidemiologistsamateur and expert sought to find something solid to help them stay afloat. Gelman points out that a pandemic can be unpredicted and impossible to predict. He says that the problem of making decisions during the pandemic was premature certainty. The ambiguity and certainty have caused many problems.

He says that letting go of the need for certainty can be liberating. This is partly where engineering perspectives come in.

A tinkering mindset

Seth Guikema is co-director of Center for Risk Analysis and Informed Decision Engineering, University of Michigan. He also collaborates with Zelners on other projects. Guikema says that if uncertainty alters the best decisions or the good ones, then it's critical to understand. It doesn't have to affect my best decisions, so it is less important.

For example, increasing SARS-CoV-2 vaccination coverage is one scenario. Even though there are uncertainties about how many deaths or cases it will prevent, the fact the program is likely to lower both is enough to convince you that a large-scale vaccine program is a good idea.

Engineers are always looking to improve their image. Susan Holmes

Holmes points out that engineers are skilled at breaking down problems into their most important pieces, using carefully chosen tools and optimizing solutions within constraints. A team of engineers is responsible for building a bridge. There are specialists in cement, steel, wind engineering, and structural engineers. She says that all the specialties work together.

Zelner learned the concept of epidemiology from his father, who was a mechanical engineer and started his own company that designed health-care facilities. His engineering background is rooted in his childhood of building and fixing things. He can tinker with a transmission model to respond to a moving target, for example.

These problems often require iterative solutions. You make changes based on what works and what doesn't. As more data is collected, you keep updating your work and learning from your mistakes and successes. This approach is very different and better suited for the complex, non-stationary issues that define public healthcare. It's also more practical than the static image that many people have of academic sciences. Here, you have a big idea, test your hypothesis, and your results are preserved in amber for ever.