GEOINT is ubiquitous, informing decision-making in almost all facets of professional and personal life. On the USGIF GEOConnect Series Main Stage Dec. 16, panelists shared how an interdisciplinary team is using Activity-Based Intelligence (ABI) concepts to combat the pandemic and specifically how the Data Pandemos project measured close physical contact of mobile devices in geographic areas to predict COVID-19 cases and inform life-saving policy decisions.
Lianne Kennedy-Boudali, Chief Strategy Officer, Concentric Advisors, moderated the conversation with Jacqueline Barbieri, Chief Executive Officer (CEO) and founder, Whitespace; Forrest Crawford, associate professor of biostatistics, Yale University; and Sid Mansur, CEO, Sentrana. The following are just a handful of the questions and answers discussed during the event.
Kennedy-Boudali: How did ABI inspire what became the Data Pandemos project?
Barbieri: The inspiration behind that was very personal. In March, I started to wrap my head around what the impact of this pandemic could be. And being a career intel analyst, I really wanted to understand what this would mean, not just for us as a country but globally. And before the vaccine, social distancing was the only tool we had. At the time it occurred to me, we know ABI, which means we can use geospatial big data and remote sensing data to measure on a real granular scale where people are and aren’t complying with that guidance. If we did that, we could possibly help people be better at social distancing and inform policymakers where people are complying with this guideline. Those thoughts led our team on a journey that involved the opportunity to meet many really brilliant and purely motivated individuals.
Kennedy-Boudali: Can you talk a little bit about the scientific contribution that this kind of data represents?
Crawford: In the context of the COVID-19 pandemic, decision-makers, especially in state and federal government, have access to a tremendous amount of data about the bad things that happen to people from COVID-19. But all of those metrics as [Barbieri] said lag actual transmission of disease by days or weeks. Therefore, by the time you know for example about an influx of ill patients to the hospital, there’s not much that a policymaker can do to save them or to prevent that from happening—it already happened. Decision-makers want to be able to prevent the precursor to those bad events, transmission of infection. So we were working with data from the state of Connecticut. They told us about all the things that happened after transmission occurred, but what we really needed was information about what was happening to the left of transmission. And how does transmission of disease happen? It happens when people are close together, within two meters. We just had no way of measuring actual human behavior relevant to the transmission of infectious disease. It was the main thing that was missing from all of our metrics, information, and projections we gave to the state government. We knew we needed to find some way of measuring this. I met [Barbieri] in late March and we discussed ways of using mobile device data to track close interpersonal contact of individuals in the state of Connecticut. We developed this probabilistic measure of close contact that could be implemented to track these close contact events statewide and it’s really different from all the other data sources that policymakers in Connecticut, nationwide, and possibly the world have access to. This guides the state’s response from targeting social distancing messaging to predicting new outbreaks, and it has become a really indispensable tool for the state because it measures behavior at the individual level in a way that we never had access to before. And it’s the exact behavior that is relevant for transmission of infectious disease.
Kennedy-Boudali: Tell us about your initial thoughts when [Barbieri] told you about this project. What made you want to get involved? And what was technically interesting and challenging to you about engaging in this work?
Mansur: My immediate reaction was that this is incredibly bold and ambitious. [Crawford] spoke to it from the scientific and epidemiological perspective, but what hit me pretty vigorously was just the computational aspect of this. What makes this such a profound challenge is that we are simultaneously dealing with incredible levels of granularity and at the same time incredible levels of scale. What do I mean by that? Granularity in the sense that we want to measure at a very microscopic geospatial level. And we want to be able to now produce insights that come from those microscopic measurements in a macroscopic way—at the level of a census block group, zip code, county, state, or country. Going from that level of granularity to that level of scale is not a trivial problem. And that goes to the heart of what we grapple with today when it comes to ABI and geospatial analytics. Geospatial analytics has effectively become micro-spatial analytics. So we need to reconcile two very diverse scales here, micro-scale and macro-scale. And we have to reconcile that very quickly with terabytes of data and make no mistake in terms of the way that data is reconciled because there is a lot that is at stake right now. Policy decisions and how we comport ourselves in terms of our response to this pandemic are going to materialize from how we reconcile these two things.