JMIR Publications

ABSTRACT

Background:

Artificial Intelligence (AI) methods offer a valuable complementary approach to public health emergency management, supporting prediction, rapid threat identification, and timely decision-making alongside the already established human-led systems and processes. However, updated and comprehensive evidence on the extent and characteristics of AI use in public health emergencies, with a specific focus on infectious hazards, remains limited globally.

Objective:

This review aimed to map the scope, nature, and extent of AI applications in public health emergency management resulting from infectious hazards, and to characterize key implementation features.

Methods:

A scoping review was conducted following the Arksey and O’Malley framework, and a search was performed in three electronic databases, including PubMed, Scopus, and the IEEE Xplore Digital Library. The search period covered studies published between January 2014 and June 2024.

Results:

A total of 613 studies were included, of which 526 (85.8%) were AI-related methodological studies and 87 (14.2%) were reviews or other article types. Across these studies, 665 infectious-hazard records were extracted, with COVID-19 accounting for the majority (387, 58.2%), followed by influenza (8.4%), dengue (4.1%), and malaria (3.0%). Publications increased steadily from 2014 to mid-2024, with a sharp rise beginning in 2019 and peaking in 2022, aligning with the COVID-19 pandemic. Notably, studies on non–COVID-19 hazards also grew between 2019 and 2023, suggesting expanding AI applications. Among methodological studies, 31.4% used social media data, mainly from X (formerly Twitter) and Weibo. Most focused on predictive analytics and disease surveillance (58.2%), followed by risk communication (23.2%), compliance with public health measures (12.5%), and policy evaluation (6.1%). Data were predominantly sourced from the USA (12.1%) and China (10.5%), with limited representation from Africa, Central Asia, and the Middle East. Funding was mainly reported from organizations in China (14.4%) and the USA (14.1%), followed by Saudi Arabia and South Korea.

Conclusions:

The findings indicate that applications of AI in infectious disease emergencies are predominantly focused on predictive modeling and surveillance, with a considerable reliance on social media data. The United States and China emerge as the primary contributors, both as sources of data and as leading funders of this research. To promote more equitable and effective use of AI in public health emergencies, there is a critical need for increased investment in local expertise, data infrastructure, and operational capacity, particularly in low- and middle-income countries.