A Student's Capstone Project That Could Inform Government Policy on Coronavirus

April 15, 2020
Katie Stanworth

NYU Abu Dhabi Senior Anh Mai is experiencing an unprecedented end to his final year as an undergraduate. As he and his classmates adjust to remote learning, isolating from one another, and contemplating the first ever virtual Commencement, Anh is slightly more informed than most when it comes to the Covid-19 pandemic.

A Major in Computer Science and Mathematics, Anh started his Capstone research in Fall 2018 and little did he know back then the relevance that it would come to have today, as he prepares to hand in his final thesis.

Entitled Policy Optimization in An Epidemic, Anh’s preemptive project aims to help government health organizations make informed decisions and implement the most effective intervention strategies in response to the outbreak of a widespread disease.

"During an epidemic, such as the one we’re experiencing so acutely right now, governments must decide on an appropriate set of actions to take in order to limit the spread of the disease, whilst also limiting the direct and societal costs of these actions. Crucially, governments often have to solve this "policy optimization" problem with only partial or limited information available,” explains Anh.

Anh’s work proposes a unique, two-phase, reactive solution to the policy optimization problem. The first phase, the Estimation Phase, proposes a unified way of modelling the flow of information from the population to the policy makers. The second, the Optimization Phase, is what sets this research apart from previous studies.

“Most epidemic papers predefine sequences of interventions and run simulations to figure out the best sequence to minimize cost,” explains Anh. “In this paper, we simulate the effect of a multitude of interventions across specific demographics to select the most cost-effective intervention policy. Our research also explores how to scale policy optimization solutions without compromising their accuracy and quality.”

Anh was first inspired to begin his research project during a study away semester at NYU New York. In a Heuristics graduate class taught by Professor Dennis Shasha, he encountered a problem in a virtual suppressor and spreader game - a simplified, simulated version of an epidemic scenario. Inquisitive and keen to learn more, Professor Shasha and Professor Azza Abouzied offered Anh the opportunity to explore the topic further as part of his Capstone.

Assistant Professor of Computer Science Azza Abouzied is Principle Investigator of the Human Data Interaction Lab at NYU Abu Dhabi. Her research team have been exploring this field for quite some time.

 

In our lab, we have been researching how to best support decision-makers who are grappling with uncertain data. For example, how to select optimal investment portfolios when we are uncertain about the future of different stocks. We then started exploring how to support decision-makers handling annual flu outbreaks who have a multitude of interventions available such as vaccinations, treatments, isolation, etc., but are unsure how to balance their cost-effectiveness. A novel epidemic, especially one like COVID-19, comes initially with a lot of uncertainty and it is important to help decision-makers select policies that save the most lives.

Azza Abouzied, Assistant Professor of Computer Science

At a time when this type of research could not be more relevant or required, Anh hopes to continue his work at NYU Abu Dhabi after graduation, and has applied for a one-year research fellowship in Azza’s laboratory.

“I hope to continue with this project by identifying and categorizing different classes of epidemic simulator. My ultimate goal is to provide a readily accessible and reliable simulator that decision-makers can experiment with in response to an epidemic.”