Eighteenth International Conference on Interdisciplinary Social Sciences

Abstract

This paper describes how in-depth interview data was used to develop an AI-based simulator that can be used to predict COVID spread under different interventions and their impact on underprivileged social groups. This simulator can simulate different environments, such as educational zones, industrial zones, fishery zones, residential zones, etc. In order to obtain data to be fed into the simulator, in-depth interviews were conducted with 189 interviewers. The sample is purposively selected to cover different occupations such as teachers, drivers, office workers, factory workers, garment workers, school students, housewives, daily wage earners, farmers, and medical professionals. Based on the in-depth interview data, a location matrix in binary form for each individual was developed that has an hourly resolution. Using these location matrices, for each occupancy, a probability density function was derived for a given location. This information was used inside the AI simulator to determine the probability of contact tracing and then to predict COVID spread. The simulator can easily be adopted for other pandemics due to its modular nature.