Inside Precision Medicine sits down with Kaitlyn Johnson, PhD, senior data analyst at The Rockefeller Foundation’s Pandemic Prevention Institute (PPI), where her work leverages data analysis and modeling to provide real-time guidance to individuals and decision-makers to prevent and mitigate pandemics. Johnson is an interdisciplinary researcher passionate about developing quantitative solutions to improve public health and medicine.
Johnson completed her PhD in biomedical engineering at the University of Texas at Austin in April 2020, just as the COVID-19 pandemic was taking off. During her graduate work, she worked in a systems biology lab that developed genomic-based tools to better understand treatment of cancer cells. More specifically, they developed technologies to track cancer cell lineages, developing linkages between genotype and phenotype.
Johnson worked on the data analysis side—integrating the outputs from sequencing data with longitudinal data of cancer cell populations over time to understand how they responded to drugs and developed chemotherapy resistance.
She started working at the UT COVID-19 Modeling Consortium led by Lauren Ancel Meyers, PhD, for her postdoc, using models of infectious disease dynamics helping to provide situational awareness and scenario-based projections to help guide the pandemic response for UT and the city of Austin.
She joined PPI last year to build upon her previous work, and to develop it into tools so that others—outside of close-knit collaborators—could leverage it. Her goal was to combine the science with the product and technology side of the world—to make tools from some of the academic science she was immersed in.
We asked Johnson about her work, the PPI, COVID-19, and future pandemics.
This interview has been edited for clarity and length.
LeMieux: I know that your work revolves around data and decision making. Can you explain it in a bit more detail?
Johnson: Let me start by saying that we’re not developing policy. What we’re doing is analysis to help inform policy. That said, we are always considering policy in our work.
When I’m conceptualizing a data to action pipeline, I think of this question in terms of three different buckets.
The first bucket is “now casting” or understanding the current state. What state are we in and what are we dealing with? In the example of a new emerging pathogen, this includes questions such as, what is the basic reproductive number (Rº) of that pathogen? What is the effectiveness of vaccination? What are these properties that help us to be able to answer the policy questions? We need to have a baseline to dive into effects of policy on the epidemic context.
The second bucket is forecasting. Based on current trends, what do we think is going to happen? In pandemic forecasting, this is challenging because it is affected by human behavior, and it is hard to predict human behavior. So, forecasting tends to be only within the next two to three weeks—what do we think is going to happen based on what the data are telling us?
The last bucket, which is more on the side of informing policy, is the decision-making part of our analysis. This is where we might make scenario-based projections to assess the effect of different policies, where we try and mathematize what a policy is. One example would be analyzing different vaccine allocation strategies based on age, geography, or other factors— and the effects of the speed and timing of those rollouts on health outcomes. Another example, and something that was done at the PPI, was to analyze the difference between requiring COVID-19 rapid tests or COVID-19 vaccines at an event, to see what mitigation measures, or combination of measures, are better at preventing event attendees from arriving infected.
Again, we are not saying, this is what you should do. We’re saying, here is the evidence to empower you to decide.
LeMieux: How is your work communicated so that it can be implemented? How do you close that gap?
Johnson: In some cases, after we have built a tool, we write an accompanying blog post. In there, we can clarify what we recommend based on the analysis. Like writing a scientific paper… the results are separate from the interpretation. Both parts can empower the public to have the evidence basis to make the informed decision.
Having the quantitative evidence, we think, can enable someone to point to that tool when talking to friends and family. They can say, this is what the analysis shows and because of this, I’m going ask you to take a test before you come over to my holiday dinner.
LeMieux: How does the PPI communicate your findings?
Johnson: That is one of the major challenges that we face as a broader public health community. We must make these tools so useful that everyone is willing and wants to use them. There’s always a tension between the complexity of the scientific message and the need for something that is easily communicated and interpretable.
One of the analogies that we think about is the weather app. People use it daily; it helps guide their daily decision-making. Because people are so reliant on weather forecasts, it encourages the collection and submission of data to the system. The idea that we have, as an Institute, is to have that type of desire for these tools within both the public and among decision makers. And that requires us working with them, answering the questions that they are most interested in, and presenting it all in a user-friendly way.
We think a lot about how to make our findings interpretable to the public so that they will be more easily accessed and how we can make them more widely available to people across the globe. It will be more impactful with a wider reach.
LeMieux: How do you feel when the evidence-based measures that you recommend are not implemented, or totally disregarded?
Johnson: What motivates me, and keeps me going, is the science.
The challenge is finding how we can meet the person or the community that we’re trying to serve where they are, and how to figure out the key questions that they’re interested in. Instead of sitting at our desk and presuming we know what people need.
For example, I did a lot of work in my postdoc supporting university policies. We found that the university had differing concerns outside of the realms of a pure health perspective. We were focused on the level of infection while they were more interested in absenteeism, quarantine time, and the costs of testing that we had not originally considered. So, we incorporated those into our analysis.
LeMieux: The pandemic is changing all the time. How do you factor in the constant change?
Johnson: The PPI has partnerships that allow us to work directly with people. And we are looking to develop tools for not just high-income countries, but also lower- and middle-income countries. We need to figure out what are the questions that they need answered? What are the types of tools that they’re looking for? We must make sure that the policy options that we are trying to mathematize, and demonstrate, are relevant to them. For example, emphasizing COVID-19 testing is not useful if tests are not readily available. So, we need to think about other policy options that we can think about incorporating. And when their needs and questions change, so too must our methods adapt to reflect this.
LeMieux: What have you learned over the past year? And what do you think about, going forward?
Johnson: We have learned the importance of interdisciplinary collaboration across sectors. That is, taking the pandemic science of academia, incorporating cost effectiveness and economic modeling, putting the communications aspect at the forefront, adding in the user center design, and then making that all into something that is production ready and scalable.
Because SARS-CoV-2 was a novel virus, those tools were not in place at the start of the pandemic. What we are trying to do is to set up those systems for currently circulating
pathogens and for novel pathogens. We want to be able to pipe in new data, answer those questions quickly, and then communicate quickly.
In general, the scientific community struggles to communicate uncertainty. So, we need to be able to say that this is what we think might happen, based on these sets of assumptions, and with a lot of uncertainty in how this could play out. Because all of it is dynamic and constantly changing. Being adaptable, while also being clear on your message, is important.
LeMieux: Do you anticipate a break after COVID-19? Or do you prepare for one new emerging pathogen after another, going forward?
Johnson: There are so many other pathogens, including endemic viruses, that could use better response tools. We’re hoping to build from what is currently circulating to be able to respond to, and have early detection of, new pathogens too.
That is some of the work that we’ve been doing in wastewater surveillance. It is multi-pathogen testing so that we could have a better idea of the baseline levels of these circulating pathogens from wastewater, and readily modify them for novel or re-emerging pathogens.
LeMieux: What keeps you up at night?
Johnson: The effect of climate change and its interaction with zoonotic spillover. In the past, we had a major respiratory pandemic every hundred years. But will that rate accelerate? And will we start to see more novel emerging pathogens because of changes in the environment that are driving movement of species into different regions?
LeMieux: I just read a Nature paper on exactly this. It is entitled, “Climate change increases cross-species viral transmission risk.”
Johnson: One of the other data analysts on our team worked closely with that first author (Carlson et al.). We’ve been thinking about it a lot. What keeps me up at night is how we are going to handle this moving forward as a society.
Julianna LeMieux, has been a science writer at GEN/IPM for four years where she covers synthetic biology genomics infectious disease, genome editing, and more. Previously, she spent years training at the bench while studying pathogenic bacteria during her PhD and postdoc. She has a passion for explaining complicated scientific concepts to a wide variety of audiences.