A major problem with COVID-19 is that even if a person is only slightly sick, they are still contagious, sometimes extremely so. This issue is magnified by the fact that people who are not aware that they’re sick, or aren’t overly sick are still able to move around and go about their daily lives.
“SARS and Ebola were more serious in the fact that if you were infected you got really sick, restricting your movement, and possibly leading to death,” said Paré (Ph.D., ’18), currently a postdoc with the Division of Decision and Control Systems at KTH Royal Institute of Technology in Stockholm, Sweden. “When you’re able to move around while only a little sick or not even realizing you’re sick, but still being contagious, you can infect more vulnerable people, resulting in a higher total number of fatalities. It’s pretty terrifying.”
COVID-19’s reproduction number is greater than 1, meaning for every person that has it, more than one other person gets sick on average. This is why cases of the virus are increasing exponentially, rather than at a steady pace. Incorporating the underlying network the propagation of viruses is one of the ways Paré has been able to develop models that can predict their spread and community impact. Paré, along with Angelia Nedić and CSL’s Carolyn Beck and Tamer Başar, have used dynamic models that capture a patient’s movements throughout a virus outbreak, unlike previous models that kept a person’s position static.
This new method allows them to model the spread of a disease more accurately, and understand and quantify how different distancing methods could work to help prevent spread. When COVID-19 data first started coming out of China, Paré was among the first to process the available data, the results of which he was scheduled to present at the Conference on Information Sciences and Systems (CISS) at Princeton University this week, before the event was canceled because of the pandemic.
Paré’s previous results from his Ph.D. research on susceptible-infected-susceptible (SIS) models may provide insight into this pandemic and the possible effectiveness of different mitigation efforts, even though COVID-19 doesn’t appear to be an SIS contagion.
After the dots (each representing a person) have had a chance to bounce around for 50-time-steps, imitating regular life, those who are sick are quarantined to the top and right borders. The two populations of sick and healthy are then allowed to move around amongst themselves, but not to intermingle. While this may be a good short-term solution, it’s not perfect.
“You can tell the sick people to stay in this area and we’ll save everybody else but the problem with that is the people in the quarantine are getting sicker because they’re interacting with each other and if you really want to save them, as soon as they’re completely healed you have to let them back into the main population before they get infected again,” said Paré. “This is hard because if you do it too soon and they’re a little sick they could affect the whole population but if you do it too late they could become sick again.”
Rather than separating the sick and the healthy, another option is to separate everyone; this is the social distancing that has been talked about, which Paré has previously called the Flee Algorithm.
“This algorithm pretty much forces everyone to avoid large groupings; change your movements completely so you’re as far away from those around you as possible,” said Paré, who will join Purdue University as an assistant professor in the School of Electrical and Computer Engineering in the fall. “This is quite difficult to implement because if you have someone who is stubborn and resisting the order they will still be moving around and potentially infecting different parts of the population.”
While this method is not perfect, Paré says it may offer the best strategy for containing the spread of the virus.
“It may be really obvious, but, given that we don’t have a vaccine yet, social distancing is the best way to stop the outbreak at this point,” said Paré.
The research discussed in this article can be found in several peer-reviewed journal articles including Analysis, Identification, and Validation of Discrete-Time Epidemic Processes and Epidemic Processes over Time-Varying Networks. Beck is an industrial & enterprise systems engineering (ISE) professor, Başar is a Swanlund Endowed Chair and electrical and computer engineering professor, and Nedić was a faculty member at ISE and is now a professor at Arizona State University.