How is artificial intelligence used in COVID-19 research?

By Tarun Sai LomteAug 12 2022Reviewed by Aimee Molineux

A recent study published in IEEE Intelligent Systems discussed the role of artificial intelligence (AI) in combating the coronavirus disease 2019 (COVID-19) pandemic.

Background

The COVID-19 pandemic has reshaped the world in an unprecedented way, resulting in more than 583 million cases and six million deaths to date. Yet, there is no clear sign of an end to the ongoing crisis. AI has been instrumental during the pandemic in supporting telemedicine, communications, automated, virtual, and economic activities.

AI has been at the center of the fight against COVID-19 from detecting severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative agent, identifying COVID-19 symptoms to saving lives and curtailing the spread of the virus. Out of over 305,900 COVID-19-related manuscripts, including preprints, from Web of Science, medical repositories, and SSRN until December 13, 2021, 38,730 were related to AI.

In the present work, the author discussed the roles of AI in COVID-19. The COVID-19 pandemic poses significant challenges to AI research, which emanate from 1) the complexity of the virus, disease, and associated data and 2) the challenges of AI tasks and processes. COVID-19 and SARS-CoV-2 present general biological system features, including interactions, self-organization, and evolution.

Challenges with AI

At the systemic and epidemic level, the pandemic is a complex open system with general and specific system complexities, which include openness, hierarchy, self-organization, interactions, heterogeneity, and dynamics. Exploring the virus and disease from different perspectives (virologic, biologic, epidemic, and medical) may help to identify the specific and holistic features.

AI systems and tasks need to address the complexities of SARS-CoV-2, COVID-19, and associated data, behaviors, processes, and systems. The corresponding challenges include 1) quantification of data complexities, 2) management of SARS-CoV-2 and disease complexities, 3) managing pandemic-related complexities, and 4) designing innovative and intelligent products, applications, and services to support testing, treatment, epidemic management, and anti-COVID-19 logistic and resource planning.

Contributions of AI to COVID-19 research

The core issues related to SARS-CoV-2 where AI has played a critical role include virus diagnosis, mutation analysis, resurgence estimation/prediction, biomedical analysis, tracing and containment, vaccine development, and systems/applications for containing the virus.

In terms of the disease, AI has been involved in pathological processing, genomic analysis, patient hospitalization, healthcare, drug development, and corresponding health/medical systems and applications. AI techniques have contributed substantially to the diagnosis, treatment, and understanding of the disease and the virus.

Nonetheless, current research has uncovered gaps and pitfalls of AI. AI research on COVID-19 has been overwhelmed by simple AI techniques, and limited research is devoted to developing original/novel AI techniques. AI research has made little progress in developing anti-SARS-CoV-2 drugs and vaccines. Moreover, cross-disciplinary AI research on COVID-19 has been limited.

AI for future pandemic management

A critical task for AI-enabled future pandemic management would be enforcing international and interdisciplinary collaborations. Pandemic management involves critical strategies such as preparation, prevention, intervention, policy-making, and mitigation efforts.

Such management is more challenging than ordinary enterprise, state-level, or multination management. As such, the author discussed different perspectives focusing on global collaborations and intelligent management that present a view of or warrant the development of epidemic or pandemic AI.  

AI for preparation and prevention of future pandemics

Multiple strategies could be enforced for preparation/warning of future epidemics or pandemics. These include AI-enabled epidemic/pandemic education and consultancy services, i.e., mobile apps, websites, chatbots, knowledge portals, and forums to address community questions. Developing protocols and policies for pandemic preparation could also be viable. AI could collect and assess protocols and historical experiences and provide feedback.

Rapid similarity analysis could be performed using AI techniques when a novel virus emerges, which could help policymakers. Quick actions could be initiated whenever new cases arise to record outbreaks, collect clinical data, share diagnostic results, and promptly report and update transmission and case data.

The development of early indicators or warning systems is an essential lesson from the COVID-19 pandemic and will be a priority for international bodies such as the World Health Organization (WHO). Enforcing global protocols and policies for such global (early) warning systems will be essential.

AI could be helpful in the automated collection, processing, and analysis of the event-based, image, textual, transactional, and medical data from public resources and enable data logging and matching programs. AI techniques could predict the timing, probability, location, and severity of an epidemic. It could also identify and assess unfair, unsafe, biased, unaccountable, and privacy-violating policies and practices and allow ethical pandemic management focusing on humanity rather than political goals.

Concluding remark

Notwithstanding the progress in AI research on COVID-19, the work is observational and shallow. Preparations for future pandemic management involve multiple opportunities and challenges, such as developing AI specific to epidemics and AI-enabled global research and developing early warning systems for pandemics.