JMIR Publications

ABSTRACT

Background:

Outbreaks of emerging and re-emerging infectious diseases keep growing in frequency and transmission rate. Healthcare facilities deliver diagnosis and treatment for confirmed cases, and they also serve as the first entry point for patients who have fever of unknown origin, pneumonia, severe infections and cluster syndromes. Researchers can use artificial intelligence to translate individual clinical observations into early public health warnings for populations, and this topic has become a core problem within smart healthcare and public health.

Objective:

This scoping review explores the technical evolution, applicable tasks and maturity of supporting evidence for infectious disease-related intelligent agents. It also examines data resources and infrastructure requirements that support the full clinical-public health task chain.

Methods:

This work used a scoping review design. It followed the JBI scoping review methodology and complied with PRISMA-ScR reporting standards. The search covers PubMed, Web of Science, Scopus and EBSCO Academic Search Complete, from January 1, 2016 up to August 8, 2026. Two separate search routes were applied: one targets intelligent agent systems, while the other covers supporting data resources. One reviewer screened all retrieved records against pre-set eligibility criteria in Rayyan, and extracted data using a piloted data charting form and all included records were subsequently rechecked for eligibility and coding consistency. The final dataset included 20 studies about infectious disease intelligent agents, alongside 59 papers focused on data resources, data engineering, data quality, interoperability, data sharing and data governance.

Results:

Agent-related research has shown a shift from rule and knowledge-driven frameworks to data-driven prediction models. Research on conversational interaction and multi-agent collaboration remain limited and lacks clinical evidence. Data-focused research has led to rich discoveries in monitoring databases, laboratory datasets, electronic health records and data cleaning workflows. However, obvious shortcomings remain in semantic interoperability, source traceability, sample representation, algorithm fairness and post-implementation governance. Reliable operation of infectious disease agents requires high-quality datasets, constrained system architectures and end-to-end evaluation systems to work together. Future research should go beyond performance testing of single-point model and conduct real-world validation in diverse institutions, tasks and time periods. It should also include data quality institutions, dynamic calibration, human coverage, accountability tracking and implementation costs within a unified assessment framework.

Conclusions:

Existing evidence shows the feasibility of independent functions in decision support, prediction, interaction and data integration. However, existing studies have not been able to confirm the safety or effectiveness of end-to-end agents that connect clinical and public health workflows. Therefore, this study proposes a constrained hybrid architecture and a phased validation roadmap as core conceptual priorities for future prospective assessments. Clinical Trial: Not applicable