COVID-19 broke out in Wuhan, Hubei Province, China in December 2019 [1], spreading across the world and, as of May 2020, claiming the lives of more than 330,000 people [2]. Caused by SARS-CoV-2, COVID-19 was declared a global pandemic by the World Health Organization in March 2020 [3]. Many individuals infected with COVID-19 experienced fever, dry cough, and fatigue; some faced a severe course of the medical condition, often requiring intensive care, including mechanical ventilation [4]. The contagious COVID-19 and its unprecedented volume of cases around the world have caused national and international disruptions to business, health care, education, transportation, and nearly every aspect of our daily lives [5]. Prompt and effective countermeasures are necessary to cap off the effects of this pandemic; comprehensive public health strategies that involve surveillance, diagnostics, clinical treatment, and research are required [6].

Leveraging digital tools and technologies to combat COVID-19 can augment public health strategies [7], for example, by using chatbots to address public inquiries about COVID-19. Additionally, using digital tools, public health professionals can track in real time the incidence of COVID-19 infections and potentially model their projection. Among such tools is artificial intelligence (AI)—a branch of computer science concerned with intelligently analyzing and handling complex information [8,9]—amplifying public health efforts against COVID-19. Despite the enthusiasm for AI applications since the 1950s, only recently have we witnessed interest in AI due to the availability of high-performance computing and vast amounts of data being generated every second [10].

AI enables machines to become intelligent, understand queries, sift through and connect mountains of data points, and draw actionable conclusions [11]. Although defining the taxonomy of AI is not trivial, its methods can be categorized based on the objective pursued: learn from knowledge, explore and discover knowledge, extract conclusions, and reason from knowledge [8].

Soon after the COVID-19 pandemic spread across the world, several governments, research institutes, and technology companies have issued calls to action urging researchers to develop AI applications to assist with COVID-19–related research [12]. From a hierarchical perspective, AI can support COVID-19 at different levels: the molecular level (eg, drug and vaccine discovery), patient level (eg, patient diagnosis), and population level (eg, epidemiological surveillance) [13].

A full review of the AI field is beyond the scope of this review, and we refer the reader to some surveys (eg, [14]) and lectures (eg, [15,16]). However, we provide a compact overview of the AI-based techniques occurring most frequently in included studies in Multimedia Appendix 1.

AI has the ability to analyze big data sets through aggregating and sifting through mountains of health care data (including patient data) to generate insights that can enable predictive analysis. The quick ability to obtain these insights helps clinicians as well as other stakeholders in the health care ecosystem to make effective, safe, and timely decisions to better serve patients and public health policy. There has been a steady rise in the number of studies regarding the use of AI techniques to resolve or address the COVID-19 pandemic [13]. Much of the AI research effort during the COVID-19 pandemic has been scattered, and a need to explore and summarize how AI technologies are being used to resolve or address the many challenges relating to COVID-19 can help us plan on how to leverage AI technologies in the current or a future pandemic. Several reviews have been conducted on AI techniques used to address the COVID-19 pandemic [12,13,17-20]. However, much of the work has been in the form of literature reviews [12,13,17-19] or systematic reviews focusing on one application of AI (eg, diagnosis and prognosis of COVID-19) [20]. Therefore, it is necessary to conduct a more systematic and comprehensive review that focuses on all applications of AI used amid the COVID-19 pandemic. Accordingly, this review aims to explore how AI technology is being used during the COVID-19 pandemic as reported in the literature. The results can be useful for health care professionals and policy makers considering leveraging AI to complement public health efforts in response to COVID-19.

In this review, we focused on any AI-based technology or approach used for any purpose related to the COVID-19 pandemic, such as diagnosis, epidemiological predictions, treatment and vaccine discovery, and prediction of patient outcomes. However, we excluded studies providing an overview or proposing a potential AI technique for COVID-19, or studies that were purely discussed from a research perspective.

We considered studies published in English between December 25, 2019, and April 12, 2020, such as peer-reviewed articles, theses, dissertations, conference proceedings, and preprints, while excluding other publications such as reviews, conference abstracts, proposals, editorials, and commentaries. We did not enforce any restrictions on the country of publication, study design, comparator, and outcomes.

Two reviewers, namely, authors AAA and MA, independently screened the titles and abstracts of the identified studies. They independently read the full text of studies that passed the title and abstract screening. We then investigated any disagreement between AAA and MA and resolved them through discussion and consensus. We calculated Cohen kappa [22] to measure the reviewer’s agreement and found it to be 0.83 for the title and abstract screening and 0.94 for the full-text reading, indicating a very good agreement [23]. Multimedia Appendix 3 shows a matrix of interrater agreement in each step.

Multimedia Appendix 4 shows a purpose-built data extraction form, which was pilot-tested using 7 relevant studies to accurately extract data. The two reviewers (AAA and MA) independently extracted data related to characteristics of the included studies, AI techniques, and data sets used for the development and validation of AI models. Like the study selection process, any disagreement between the reviewers was resolved through consensus. We calculated Cohen kappa [22] and found it to be 0.88, meaning a very good agreement [23].

After extracting the data from the identified studies, we used a narrative approach to synthesize it. Specifically, we classified and described AI techniques used in the included studies in terms of their purposes (eg, diagnosis and drug and vaccine development), AI area or branch (eg, traditional machine learning and deep learning), AI models and algorithms (eg, decision tree, random forest, and naive Bayes), and platform (ie, computer and mobile). Further, we described the data sets used for development and validation of AI models in terms of data sources (eg, public databases and clinical settings); type of data (eg, radiology images, biological data, and laboratory data); size of the data set; type of validation; and proportion of training, validation, and test data sets. We used Microsoft Excel (Microsoft Corporation) to manage data synthesis.

We retrieved 435 studies through searching the identified bibliographic databases (Figure 1). Of those studies, we removed 53 duplicates; we then screened the titles and abstracts of the remaining 382 studies. The screening process led to the exclusion of 234 studies for reasons detailed in Figure 1. After reading the full texts of the remaining 148 studies, we excluded 73 studies, as they did not meet all eligibility criteria. Thus, we included the remaining 75 studies. We identified 7 additional studies by checking reference lists of the included studies and relevant literature reviews. Overall, 82 studies were included in this review.

Among the included studies, 72 were preprints and 10 were published articles in peer-reviewed journals (Table 1 and Figure 2). About two-thirds (n=53) of the included studies were submitted in March 2020, and the remaining studies were published in February and April 2020. However, no studies were published during the first 2 months of the COVID-19 outbreak. The included studies were conducted in 19 countries; however, half of the studies (n=41) were published in China. Multimedia Appendix 5 shows the characteristics of each included study.

In this study, we conducted a scoping review of the use of AI against COVID-19. We found a lack of publications in December 2019 and January 2020. This is not surprising, given that SARS-CoV-2 was only identified on January 7 [106]; insufficient data was not available to back scientific publications, in particular internationally; and the contagiousness and aggressiveness of the virus were underestimated (first lockdown in China was January 23 [106]). Half of the studies in this report were published in China. Since SARS-CoV-2 originated in China and affected it the most during the first 3 months of the pandemic, it had the most data related to COVID-19. Considering lengthy publication processes and the vast number of COVID-19–related manuscript submissions, it is also not surprising that most of the included studies were preprints.

In the included studies, AI was used for five purposes: diagnosis, treatment and vaccine discovery, epidemiological modeling, patient outcome–related tasks, and infodemiology. None of the included studies used AI for other purposes such as contact tracing of the individuals, providing training to students and health care professionals, or robotics to deal with suspected and quarantined cases.

Most of the AI techniques used in the included studies were based on deep learning approaches such as CNN and RNN. All but 1 study used desktop machines, workstations, and clusters as opposed to mobile platforms. This can be explained by the computational demand in training AIs. Although all major mobile phone manufacturers equip their flagship models with AI coprocessors, these coprocessors accelerate inference, a computationally much lighter task. In addition, federated learning [107] (a machine learning privacy-preserving technique usually used in mobile phones) is still in its infancy and raises issues such as data sovereignty, scalability, and performance.

Data sources used in the included studies usually came from the public domain (eg, NCBI, GitHub, Kaggle) and proprietary databases (less common). Radiology images were the most commonly used type of data, followed by biological data. The number of samples was still comparably small (less than 1000 in half of the studies). The diversity and size of data indicate a lack of publicly available data despite COVID-19 cases having surpassed 5 million at the time of writing. We, therefore, second Wynants et al [20] call “for immediate sharing of the individual participant data from COVID-19 studies worldwide.”

Although this review explores the use of AI against COVID-19, some applications could prove useful far beyond this pandemic. For instance, Kiwibot designs autonomous medical delivery robots to minimize interpersonal contact [108]. Whiteboard Coordinator developed a high precision thermal screening device eliminating individual measurements, leading to higher throughput and larger social distances [109]. Although mobile phones are not yet the AI platform of choice, the first apps to track interpersonal contact using mobile phones have been published to prioritize COVID-19 testing [110]. Finally, whereas a real-time reverse transcription polymerase chain reaction test takes around 25 minutes and requires stocks of chemical reagents, AI can inspect chest CTs to provide preliminary diagnoses in seconds. We believe that increasing social distance and providing fully autonomous checkups will be the most valuable use of AI beyond the current pandemic.

In the past, fundamental AI research was focused mainly on faster (or even feasible) training. We believe that, in the future, this must be complemented with public education. AI mistrust, because of our still lacking understanding of how AI works at the deepest level, further raises ethical questions that need to be answered before AI will be uniformly accepted. We also found that AI features and results were reported in an inconsistent manner, potentially fueling AI mistrust and making a direct comparison between studies difficult. Of the 82 studies, we found that only 64.6% (n=53) of the studies included in this review disclosed the type of validation, 61% (n=50) mentioned the data size, and more than 7% (n=6) did not even specify the type of AI used. It is therefore important that we as a community develop a standardized reporting protocol to slow down the barrage of poorly conducted COVID-19 studies that threaten to overwhelm serious scientists (1916 related papers were retrieved before April 5, 2020 by Wynants et al [20]), strengthen properly conducted studies, and improve reproducibility.

We found that, explicably, the landscape of studies is still dominated by Chinese institutions, which bears the potential for cultural, technological, and geospatial biases. However, we see a recent move toward a more balanced landscape (see Figure 2). Although we identified more than 100 models developed in the included studies, we did not assess their quality as it was out of this review’s scope. Therefore, further reviews are needed to assess the quality of AI models used in the fight against COVID-19.

Given the current “infodemic” [13], we find it surprising that NLP is not used more often. We see AI-based analysis of effective advertisement for nonpharmaceutical interventions as one research opportunity to answer questions like what manner of speech results in maximum public acceptance.

In this study, we provide a scoping review of 82 studies on AI against COVID-19. Given that many of the proposed methods are not yet clinically accepted, we remark that the most rewarding research will be on methods promising value beyond COVID-19. We believe that mobile phones offer unexploited potential, but more research in the direction of energy-efficient and federated learning is needed. We also believe that the use of NLP to assess effective communication of nonpharmaceutical interventions is a largely unexplored research direction, especially since data driving this research is available in the public domain, unlike much of the data produced by clinical studies. For AI to gain broad acceptance, standardized reporting protocols to be followed by studies on AI are needed. Likewise, more research on AI ethics and explainable AI is needed, paired with public education initiatives.