Probability, differential equations, and catastrophe models united against Ebola

Paolo Bocchini is a an Assistant Professor of Structural Engineering at Lehigh University and Javier Buceta is an Associate Professor of Chemical Engineering at Lehigh University. Their research synergy, together with Graziano Fiorillo, a Postdocoral Research Associate in Dr. Bocchini’s and Dr. Buceta’s labs, led them to begin developing models for predicting ebola outbreaks.  

What is the chance that two structural engineers and a physicist team up to fight one of the deadliest diseases in the history of humankind? Well, it looks like the plot of a Dan Brown novel, but it really happened, and it all started literally by “chance”, probability.

In 2014 a group of us, Lehigh University faculty, noticed that our university has a high density of researchers interested in Probabilistic Modeling and its applications to engineering and science, spread across various departments and colleges. For this reason, we decided to start coordinating our graduate courses to create a better synergy. But you know how it works: if you put two or more professors in the same room, they start talking about their research. So, at some point, Javier described his innovative way to model the non-homogeneous migration of bats infected by Ebola, which seems to be the main mechanism in which the virus travels for hundreds of miles triggering outbreaks in cities that did not see it coming and are completely unprepared for it, with devastating effects. Then Paolo noticed that the mathematical formulation and the type of uncertainties in the model that Javier used for infected bat migration have strong similarities with the way in which he addresses the uncertain propagation of seismic waves over a large region. It was (scientific) love at first sight. Paolo and Javier immediately saw the potential of combining Paolo’s novel hazard models and the rigorous framework that civil engineers use for catastrophe modeling, with the cutting-edge technique that Javier was developing to capture the disease spreading. The outcome is a comprehensive tool that can predict (in a probabilistic sense) the risk of Ebola outbreaks over a region as broad as the entire African continent and, in this way, drive preemptive allocation of limited resources in the most effective way, to fight promptly outbreaks if they happen to occur.

With this idea, we (Paolo and Javier) submitted a Collaborative Research (CORE) proposal that was funded (thank you Lehigh!) and allowed us to hire a postdoc and bootstrap this new line of research. As you may imagine, it wasn’t easy to find a person with the right competences and enough curiosity to join us in this adventure at the boundary of several disciplines. Luckily, we found Graziano, who with his expertise in probabilistic modeling applied to engineering problems and his proficiency with high-performance computing has been the perfect scholar to carry on this project. With enthusiasm, our “bold trio” started working against Ebola in early 2016 (some people make fun of us saying that we are rather a “bald trio”).

Over this year, the outcome of the project has been twofold. On one hand, we had to work hard to transform an exciting idea into a real technique that can be used in practice. Despite the common background in mathematics and probability, bridging, combining, and integrating our expertise in such different fields of application has not been trivial. Nevertheless, we were able to submit two credible proposals (one to NIH and one to NSF, in collaboration with Shin-Yi Chou, who joined our team). Even though the proposals were not successful at the first round, the reviews were very encouraging and we are working on the resubmission. We have also a paper accepted for the proceedings of the most prestigious conference on applications of probability to civil engineering, and three journal manuscripts in the pipeline. On the other hand, we have considered all along our project as a pilot of what can be accomplished leveraging Lehigh’s strength in interdisciplinary probabilistic modeling. For this reason, we continued working to strengthen the Probabilistic Modeling Group, and we “went public” last October with the organization of the “Symposium on Probabilistic Modeling in Engineering and Science” at Lehigh University. The symposium was a success, with 130 registered participants from Lehigh and other universities, such as Rice, Carnegie Mellon, Johns Hopkins, and Princeton, three world-class speakers and a strong support from Lehigh Administration, with the direct participation of Vice-President Snyder, Dean DeWeerth, and Associate Dean Coulter. In particular, Dean DeWeerth suggested to change the name to “FIRST Symposium on Probabilistic Modeling in Engineering and Science” and make it a tradition of Lehigh University. The Probabilistic Modeling Group has also submitted a proposal to establish a graduate certificate in probabilistic modeling that spans across centers, departments and colleges, and which may become a MSc or PhD program down the road.

Almost every quantity in engineering and science is affected by randomness and uncertainty, to a certain extent. Sometimes, modeling such randomness and uncertainty is necessary to capture the studied phenomenon, and be able to predict its outcome (in a probabilistic sense). We believe that in the next 20 years, probabilistic modeling will become a fundamental skill for all engineers and scientists, as computational modeling has become. Our work to predict and fight Ebola outbreaks is an attempt to showcase the potential of probabilistic modeling and Lehigh’s strength in this field, while also bringing the engineering mindset of problem solving into the world of epidemiology.