DeepTrack: An ML-based Approach to Health Disparity Identification and Determinant Tracking for Improving Pandemic Health Care

Jinwei Liu, Long Cheng, Ankur Sarker, Li Yan, Richard A. Aló

Abstract: The Coronavirus disease 2019 (COVID-19) pandemic has severely impacted countries around the world with unprecedented mortality and economic devastation and has disproportionately and negatively impacted different communities—especially racial and ethnic minorities who are at a particular disadvantage. Black Americans have a long-standing history of disadvantage (e.g., long-standing disparities in health outcomes) and are in a vulnerable position to experience the impact of this pandemic. Some studies indicate high-risk and vulnerability of the elderly and patients with underlying co-morbidities, however, little research paid attention to leveraging geographic information and machine learning (ML) to track the social and structural health determinants, which can provide a lower level of granularity. In this paper, we propose DeepTrack, a geospatial and ML-based approach to identify diverse determinants (including the structural, social, and constructural determinants) of health disparities in COVID-19 pandemic, which provides a lower level of granularity. We provide a thorough analysis of health disparities and diets based on multiple COVID-19 datasets and examine the structural, social, and constructural health determinants to assist in ascertaining why disparities (in racial and ethnic minorities who are particularly disadvantaged) occur in infection and death rates due to COVID-19 pandemic. We track determinants of nutrition and obesity through diet examination. Extensive experimental results show the effectiveness of our approach. The research provides new strategies for health disparity identification and determinant tracking with a goal to improve pandemic health care.