Simulating an Epidemic: Effective Ways to Beat the Curve by Joshua van Cuyck | Somerset College
Please click here for the COVID-19: South African Government resource and news portal.

Simulating an Epidemic: Effective Ways to Beat the Curve by Joshua van Cuyck

17th September 2020

The COVID-19 storm arrived without warning and has left our government in a scramble to uphold the already-struggling healthcare system. Our leadership has not been left alone to fend in the frenzy, though, as epidemiology and data science provides a window through which we can observe the devastating effects inside a silicon-driven simulation.

 As any intellectually unstimulated and curious grade 12 student would do with their newly freed schedule, I thought it would be interesting to open this window, tinker with potential scenarios, and discover the most effective solutions to a seemingly impossible problem: How do we flatten the curve?

The rules are simple and are based on the SIR model. There are those who are Susceptible, those who are Infected and those who have Recovered. For every unit of time spent within a certain radius of an infected individual, there is some probability of contracting the virus. Physical proximity and probability act as a stand-in for the typical handshake or sneeze. This the closest one can come to simulating transmittance through the SIR model.      

These simple rules produce startling yet immensely applicable results, which, if analysed and implemented properly, can improve the social well-being and productivity of any school or business.

R0 and the Hygiene: Distance Ratio:

Spatial parameters can be tweaked to simulate the effectiveness of different prevention techniques. In epidemiology, the Basic Reproductive Number, denoted by R0, represents the expected number of cases generated by a single case in an environment where the entire population is susceptible to infection. Without preventative techniques, Ro sits around a comfortable 3.4 in this simulation.

Halving the probability of infection has the effect of halving R0. This seems intuitive, but upon implementing a central place - simulating your local grocery store - and experimenting with infection radii, decreasing the probability of visiting the store by a factor of 5 has a remarkably identical effect!

This is extremely useful information and shows us that the effect of hygiene-consciousness is ­2.5 times greater than the routine-altering and socially frustrating technique of reducing consumer market availability.

This information can be useful when deciding what preventative measures should be implemented within an office or school environment to achieve the balance of maximum productivity and minimum risk. Instead of limiting the use of meeting locations, implementing a mandatory hand sanitiser would have a greater effect without hindering productivity.

The Danger of Science Denial and the 2nd Wave:

Upon further investigation into the most effective techniques for flattening the curve, I decided to simulate the effect of social distancing and its inevitable dismissal.

The results were as follows: Immediate social distancing across 100% of the population eradicates the virus at a rate around 10 times faster than without any other measures, and the highest infected population had a similarly significant drop. This puts less pressure on the public healthcare system and is most certainly the most effective eradication technique. Immediate social distancing across 70% if the population flattens the curve but prolongs the duration significantly. This is where the problem lies.

In this circumstance, R0 < 1, indicating that the spread of the virus is no longer exponential, and daily infections are on the decline. Governments may be pressured and inclined to drop social distancing regulations. Simulating this proves to have the devastating effect of a second exponential wave of infections. Science does not lie.

A Positive Outlook:

Real epidemiology takes into account contact-tracing and isolation as well as travel between communities. Although the results produced by the limited number of rules applied to this simulation were useful, it is impossible to simulate entirely the real-world effect of different techniques, as well as the effectiveness of different techniques working in unison.

We are lucky to be facing such a pandemic during an exciting boom in data science and technology and it is interesting to observe what the silicon chip can do for us when presented with an environment with the properties of nature. I am delighted to uncover some of the secrets of flattening the curve!