How to prevent viruses from going viral?

In his previous work, Markus Sinnl was actually concerned with social media such as Twitter. In recent years, he has developed methods to understand how certain posts go viral. When the coronavirus pandemic erupted, he realised that the spread of information in social media showed great similarities with the spread of the coronavirus in the population. In a project funded by the Austrian Science Fund FWF, he now wants to employ his tools in the fight against the coronavirus.

Search for exact solutions

To this end, principal investigator Sinnl plans to expand the methods developed for social media. In his model, the virus is equated to an agent in a network who wants to achieve maximum propagation. “We would model this in exactly the same way as the spread of a message in a social network, but we want to add an extra layer consisting of measures taken from outside the network to prevent spread,” Sinnl explains.

Sinnl uses a method going by the name of integer programming, which involves the solutions of equations that do not contain numbers with fractional components, but only whole numbers. The method is well suited to handling many practical problems in industry or, for instance, public transport timetables, a fact that is corroborated by Sinnl conducting his research at the Institute of Production and Logistics Management at the Johannes Kepler University Linz. The advantage of integer programming: the equations provide more than just probabilities. “Our methods deliver what can be proved to be the optimal solution to the problem,” Sinnl notes emphatically.

Methods from advertising

“In the last few years I applied integer programming to solve influence maximisation, among other things,” he says. “The question was: how can you influence people in social networks?” Whereas such problems do have a tradition in computer science according to Sinnl, most people try to solve them “heuristically”, that is, by finding approximate solutions.

When describing the spread dynamics of the coronavirus, exact solutions would be a novelty; there, too, heuristic methods and approximate solutions have been used up to now. Although models are used to test the effectiveness of certain measures to prevent the spread of the virus, Sinnl explains that it is almost impossible to determine whether they are actually the optimal solution. Sinnl adopts another approach by looking at the situation as an optimisation problem. He maps various measures for reducing spread in his methods and then searches directly for the optimal combination.