Models for Incorporating Non-Pharmaceutical Interventions


There are a lot of questions circulating about the COVID-19 pandemic that people want answered, but the problem is we do not have the immediate answers. Governments have been reaching out to epidemiologists, medical practitioners and others who have some expertise in disease modeling in order to better understand the trajectory of the virus and its impact on society.

One of the most widely used models are called Mathematical Compartmental Models, which simplify the mathematical modeling of infectious diseases. With this basic model, the population is assigned to compartments with labels S, I or R (susceptible, infectious or recovered). This is the basis for all the models, and the letters and paths can be adapted to specific outbreaks. The labels represent the progression of individuals between compartments. For COVID-19, most people use the model SEIS, which represents susceptible, exposed, infectious then susceptible again. As we are learning more about asymptomatic carriers, this diagram better describes the potential progression of individuals through the compartments, with two different possible courses after exposure. All individuals are susceptible, and then progress to exposed. From there, individuals are either asymptomatic and then recover, or they become infected and then either recover or die.

Because we don’t yet have a vaccine, the best we can do is to use mitigation measures like case isolation, voluntary home quarantine and wearing face masks are aimed at reducing the spread of the virus as well as the demand on health facilities. For example:

Social Distancing – If you reduce social exposure by 75 percent, if one person is infected today, in five days, the infected individual will infect less than a person (0.625 people), in 30 days, the one person will infect just 2.5 people. If you reduce exposure by just 50 percent, then the one infected person can infect 1.25 people in five days, and 15 people by day 30. When there is no social distancing, one person can infect up to 406 people in 30 days.

Face Masks – If a COVID-19 carrier is not wearing a face mask and they come into contact with a healthy individual who is wearing a face mask, the transmission probability is 70 percent. If the carrier is wearing a face mask when they come into contact with a healthy individual who is not wearing a mask, the transmission probability is five percent. If both individuals are wearing a face mask, the transmission probability drops to 1.5 percent.

The models we have access to for COVID-19 produce forecasts, not projections, meaning, “They do not explicitly and effectively model the effects of interventions or other “what if” scenarios. Models do, however, assume that interventions will be implemented and will continue to be upheld in the future, resulting in an overall decrease and growth rate of the disease”[1]. Forecasts are attempts to predict what will happen versus projections, which are attempts to describe what would happen given certain hypotheses. In as much as government policies towards mitigating COVID-19 are essential, individual responsibilities of keeping him/herself safe is the most important game changer in fighting against COVID-19. 



[1] Eduado Massad et al., (2005): Forecasting versus projection models in Epidemiology: The case of the SARS epidemics. Journal of Medical hypothesis. (65) pp 17-22.

Original post