My research is primarily on the modeling and optimization of interactive decision making in real-world multi-agent systems from the perspective of operations research. In particular, I am interested in healthcare applications, where the accuracy of models and results is of vital importance. This area remains highly active, as the rapidly growing populations, along with advancements in relevant technologies and means of interactions, give rise to a continuously expanding set of open challenges and opportunities for the enhancement of healthcare services in terms of quality and effectiveness. My general approach to such problems is based on the adoption and integration of tools and methods from various disciplines, including game theory, network science, complex systems, computational optimization, artificial intelligence, and machine learning. I am particularly interested in the applications of novel theoretical and computational methods in capturing the dynamics of large systems, instances of which are leveraging inverse game theory for inference of agents’ objectives from real-world observations, and mean-field game theory for reducing the complexity of analyzing the dynamics of interactions in large multi-agent systems.
Current Projects:
Modeling the Social Response of Populations to Disease Spread using Spatial Game Theory
This research is on modeling the social response to disease spread when decision makers are not individuals, but the populations of individuals or health policy makers. To approach this problem, I developed a model based on the public goods game in the settings of a spatial game on a network. In this model, populations are considered as agents whose aim is to maximize the collective benefit to their society as well as that of their neighboring populations. In this setting, populations can adjust their impact and contribution to public health by selecting their level of vaccination. This model accounts for various factors affecting the cost assessment of these groups, including the cost of vaccination, the cost of contracting the disease, and an “awareness” factor due to the dissemination of information through public media. Furthermore, the transmission dynamics of the disease was captured within the well-known Susceptible-Infected-Resistant (SIR) model.
I am also investigating the application of machine learning models to the modeling and control of complex systems such as epidemics. Accordingly, I am currently working on developing an approach based on deep reinforcement learning to leverage the developed model and the vaccination dynamics from real-world observations in solving multivariate optimization problems in epidemic control policies.
Emergency Facility Location Problem with Prioritized Demand Groups
The problem tackled in this study is the optimal allocation of a limited number of emergency stations to different districts at the time of crisis, considering that there are different demand groups with different priorities. The specific case addressed in this study is the recent Covid-19 outbreak and the problem of opening a limited number of temporary health centers in a region according to the age distribution of people in different neighborhoods. In this study, the distribution of age is considered to affect the priority of groups as older groups are more at risk of demanding emergency healthcare. The goal of this study is to choose the location of health centers to minimize the maximum weighted distance of different neighborhoods from the health centers; making temporary health centers accessible to all groups but by prioritizing the groups which are most likely to need such centers.
To solve this problem, we propose a clustering-based approach in a p-center model. This project enables public health administrators to better decide and design relevant policies for allocation of limited healthcare resources to different populations. Furthermore, the novel approach proposed in this work extends the set of techniques currently employed to solve the facility location problem.