Sick MD: Molecular Dynamics-like simulations of infection spread

A recent article on the Washington Post showed how the spread of diseases can be modelled using Molecular Dynamics (MD), with each individual being an atom. Atoms (or people) can be assigned three distinct values – healthy, sick, and recovered. An atom that has recovered cannot get sick again, an assumption based on the fact that immunity acquired after getting sick lasts for some time if the immune system is healthy and no external antibodies were used to treat the infection. For the case of COVID-19, the disease primarily appears to spread through cough droplets, hence the spread can be modelled by physical proximity collisions between molecules. The rules for the spread are simple –

  1. Healthy atoms colliding with each other doesn’t change anything.
  2. Healthy atoms colliding with recovered atoms doesn’t change anything.
  3. Recovered atoms colliding with sick atoms doesn’t change anything.
  4. Sick atoms colliding with healthy atoms makes the healthy atoms sick.
  5. Sick atoms change to recovered atoms after a certain period of time.

Ofcourse, this simplistic model doesn’t take into account things such as asymptomatic cases,  deaths and re-infection, but it is a powerful tool to show the qualitative  effects of policy measures, such as quarantines and shutdowns on the spread of the disease. Building a more accurate and comprehensive model might be similar to our efforts at building a multiscale fluid solver : MD can give detailed information at the level of houses and streets, but more continuum methods might be fed this information to model entire cities and country.

The message that this model offers, however, is simple – stay at home, do not go out unless very essential.