In the past several months, we have seen that effectively distributing COVID-19 vaccinations is an incredibly important and daunting task. Due to a wide and asymmetric imbalance between supply and demand, finding the right balance, timing, and utilization of vaccinations has proven difficult for everyone involved. Organizations need to correctly predict demand at different times at specific locations to minimize spoilage and optimize utilization rate while accounting for limited supply.

In this talk, Samira Soleimani, Data Scientist at CGI, will demonstrate how machine learning algorithm and multi-objective optimization can be used to help tackle this challenge. First, Samira will explain the difference between single-objective and multi-objective optimizations. Second, she will introduce and compare line and curve-fitting approaches to define objective functions for distribution optimization. Then, she will explain and implement multi-objective particle swarm optimization (PSO) in the python package “PyGMO” to find the optimal point for vaccine distribution in a neighborhood.