Public Health Monitoring System In COVID-19 Conditions Using Machine Learning-Based Sentimental Analysis

Abstract

Twitter is a significant forum for individuals to discuss and disseminate health-related data.  The system offers substantial data for immediate monitoring of contagious diseases (such as COVID-19), relieving disease-prevention organizations from the laborious tasks associated with Personal Health Measures (PHM). PHM identification is a crucial technique for staying informed about the status of an epidemic. It aims to determine an individual's health by analyzing web text data. This research investigates the process of identifying PHM related to COVID-19 using data from Twitter. The research has constructed a COVID-19 PHM dataset with tweets labeled with four distinct categories of health disorders connected to COVID-19: self-mention (SM), other-mention (OM), awareness, and non-healthcare (NHC). The research achieved favorable outcomes in the PHM identification task. The categorizing results enable prompt health tracking and oversight for digital epidemiology. The study assesses the impact of the attention strategy and training methodology on the predictive capabilities.