Artificial Intelligence Applications with Covid-19 in the First Year of the Pandemic: A Systematic Review

Abstract

Background

WHO declared COVID-19 as a pandemic in March 2020. This crisis made AI intensively used to serve healthcare providers to diagnose, monitor and treat patients through AI-based solutions and applications.

Objectives

To systematically review the existing literature that applied Artificial Intelligence and Machine Learning algorithms to deal with COVID-19 during the first year of the pandemic from January 2020 until March 2021.

Methods

Data collection was performed to retrieve research studies from PubMed and Google Scholar. Studies were scrapped using the PubMed API widget in Orange. This resulted in 421 studies.  For Google Scholar, Serp API, to scrape the search results as JavaScript Object Notation (JSON) files. A simple Python program using PyCharm^[1] were used to parse the JSON and extract the required information for retrieved studies which resulted in 981 studies.

Results

Attention was focused on Prediction and Diagnosis categories. Mostly for predicting and forecasting the number of new cases; the Outbreak has gained the most attention then Predicting the severity level of Covid-19 patients with 75 and 37 published studies. In diagnosis most publications applied chest-radiography followed by clinical data subcategories. The categories of Screening, Classification of Genomes, and Prognosis have not gained much attention.

Conclusions

An accelerated growth/increase in research, and published studies concerning the employment of AI/ML technologies to deal with the Covid-19 pandemic overtime was observed. AI methods and techniques helped understand and analyze the pandemic data and prepare resilient evidence-based/driven plans and technologies to deal with COVID-19.

^[1] "PyCharm: the Python IDE for Professional Developers by JetBrains." https://www.jetbrains.com/pycharm/. Accessed 13 Dec. 2021.