Description
Programs which aim eradicate disease must rely on interpretable models. These models quickly become hard to solve, not to mention train on missing parameters. Scipy and PyMC come to our rescue for the heavy lifting.
In 2018, Israel has seen the biggest outbreak of measles since the introduction of a vaccine in the late 1960s. Nowadays, vaccine policies are not only decided by laboratory tests. Those tests are complemented by a plethora of computational epidemiology simulations predicting the effects of various vaccination policies on the entire population. A population-level policy to eradicate disease must rely on Interpretable models. These models quickly become hard to solve, not to mention train on missing parameters. Using Scipy as a solver, and PyMC for Bayesian inference we are able to learn parameter distributions for missing natural parameters, such as the disease's "strength" or "infectiousness". We can then use the underlying distributions for these parameters in order to simulate possible outcomes for future policies.