Crafting Data-Driven Strategies to Disentangle Socioeconomic Disparities from Disease Spread

Abstract

As a disease whose spread is correlated with mobility patterns of the susceptible, understanding how COVID-19 affects a population is by no means a univariate problem. Akin to other communicable diseases caused by viruses like HIV, SARS, MERS, Ebola, etc., the nuances of the socioeconomic strata of the vulnerable population are important predictors and precursors of how certain components of the society will be differentially affected by the spread of the disease. In this work, we shall delineate the use of multivariate analyses in the form of interpretable machine learning to understand the causal connection between socioeconomic disparities and the initial spread of COVID-19. We will show why this is still a concern in a developed nation like the USA with a world leading healthcare system. We will then emphasize why data quality is important for such methodologies and what a developing nation like India can do to build a framework for data-driven methods for policy building in the event of a natural crisis like the ongoing pandemic. We hope that realistic implementations of this work can lead to more insightful policies and directives based on real world statistics rather than subjective modeling of disease spread.