A hybrid contextual framework to predict severity of infectious disease: COVID-19 case study

Abstract

Infectious disease is a particular type of disorder triggered by organisms and transmitted directly or indirectly from an infected one like COVID-19. The global economy and public health are immensely affected by COVID-19, a recently emerging infectious disease. Artificial Intelligence can be helpful to predict the severity rating of COVID-19 which assists authorities to take appropriate measures to mitigate its spread in different regions, hence it results in economic reopening and reduces the degree of mortality. In this paper, a hybrid contextual framework is proposed which incorporates content embedding of Standard Operating Procedure's (SOPs) auxiliary description along with COVID-19 temporal features of the respective region as side information. The word embedding techniques are incorporated to generate distributed representation of SOPs auxiliary description. The higher representation of auxiliary description is obtained by utilizing content embedding and then combined with temporal features to build counties profiles. These county profiles are fed into a profile learner based on an ensemble algorithm to predict the severity level of COVID-19 in different regions. The proposed contextual framework is evaluated on public datasets provided by healthdata.gov and the National Centers for Environmental Information. A comparison of the proposed contextual framework with other state-of-the-art approaches has demonstrated its ability to accurately predict the severity level of COVID-19 in different regions.