Using monitoring-oriented techniques to model the spread of disease

Abstract

Disease modelling can be a very complex task since every disease is unique with its own way of spreading. There are also various ways how it can be modelled. This project considers the modelling of diseases through a technique called Monitoring-Oriented Programming (MOP) where a number of monitors are able to keep track of the persons and objects in a simulation and modify system behaviour to simulate spread from one to another accordingly. Such a simulator may become very complex to implement without MOP due to the various different factors determining transmission particularly those concerning real-time. Furthermore, having code related to the spread inlined with that driving the simulation makes it less understandable. MOP helps achieve this separation of concerns since the monitoring code which determines spread is isolated from the system and only hooks onto the simulation. However, this technique has not been applied to this problem before and hence there are concerns whether this is possible. Throughout this project, a runtime verification tool is used to implement 3 different disease spread models using 3 different computational approaches based on the SEIR compartmental model where one is either Susceptible, Exposed, Infected, Recovered or Removed at a particular time. Each model and its approach are placed in a separate script which is then compiled using the tool to generate monitors which are able to latch onto the simulation. The simulation itself consists of 3 different scenarios representing different situations in which a crowd setting is modelled. This implementation not only shows that it is possible to implement this problem using MOP, but also through the evaluations carried out it is shown that while there are monitoring overheads, the approach scales up making it a suitable solution.