Artificial Intelligence in Action: Addressing the COVID-19 Pandemic with Natural Language Processing

Area of Science:

• Artificial Intelligence
• Computational Linguistics
• Public Health Informatics

Background:

• The COVID-19 pandemic created significant societal challenges, including widespread health effects and information overload for researchers and the public.
• Effective management of the pandemic necessitates addressing critical information needs.
• Natural Language Processing (NLP) offers a powerful toolkit to tackle these challenges.

Purpose of the Study:

• To review the application of NLP techniques in addressing information needs during the COVID-19 pandemic.
• To survey existing NLP studies, systems, and datasets relevant to the pandemic.
• To identify trends and challenges in the use of NLP for public health crises.

Main Methods:

• Systematic review of approximately 150 NLP studies and over 50 systems/datasets.
• Categorization of NLP work based on core tasks: information retrieval, named entity recognition, literature-based discovery, and question answering.
• Analysis of additional NLP tasks applied to pandemic-specific issues: topic modeling, sentiment analysis, caseload forecasting, and misinformation detection.

Main Results:

• Detailed examination of NLP applications across various tasks, highlighting their utility in managing pandemic-related information.
• Identification of over 50 systems and datasets developed for COVID-19 NLP research.
• Demonstration of NLP's versatility in addressing diverse pandemic challenges, from scientific literature analysis to public sentiment monitoring.

Conclusions:

• NLP is a crucial technology for navigating the complexities of health crises like the COVID-19 pandemic.
• Continued research and development in NLP are essential for improving information access, analysis, and dissemination during public health emergencies.
• Addressing remaining challenges in NLP will further enhance its role in future pandemic response efforts.