Leveraging Machine Learning to Unveil the Critical Role of Geographic Factors in COVID-19 Mortality in Mexico

In this paper, we present an in-depth analysis leveraging several renowned machine learning techniques, including Snap Random Forest, XGBoost, Extra Trees, and Snap Decision Trees, to characterize comorbidity factors influencing the Mexican population. Distinct from existing literature, our study undertakes a comprehensive exploration of algorithms within a defined search space, conducting experiments ranging from coarse to fine granularity. This approach, coupled with machine learning -driven feature enhancement, enables us to deeply characterize the factors most significantly affecting COVID-19 mortality rates within the Mexican demographic. Contrary to other studies, which obscure the identification of primary factors for local populations, our findings reveal that geographical factors such as residence location hold greater significance than even comorbidities, indicating that socioeconomic factors play a pivotal role in the survival outcomes of the Mexican population. This research not only contributes to the targeted understanding of COVID-19 mortality drivers in Mexico but also highlights the critical influence of socioeconomic determinants, offering valuable insights for public health strategies and policy formulation.