Photo Credit: solarseven
The following is a summary of “Federated learning as a smart tool for research on infectious diseases,” published in the November 2024 issue of Infectious Disease by Zwiers et al.
The use of real-world data in infectious disease (ID) research has grown, especially since the COVID-19 pandemic, but privacy concerns and data being stored across various sources limit the ability to centralize and fully leverage the combined data.
Researchers conducted a retrospective study to explore the applications of federated learning (FL) in ID research, highlighting its opportunities and challenges in addressing privacy concerns.
They identified references for this review by searching MEDLINE/PubMed, Google Scholar, Embase, and Scopus up to July 2023. The studies were selected based on FL in various ID applications.
The results showed that 30 references were included and categorized into 4 sub-topics: disease screening, clinical outcome prediction, infection epidemiology, and vaccine research. Most of the research focused on COVID-19, FL accurately predicted diseases and outcomes, outperforming non-federated methods in all studies. However, most studies did not utilize real-world federated data, and showcased FL’s potential through manually partitioned data.
Investigators concluded FL as a promising methodology for ID research, but further exploration of its applications was required to realize its full potential.
Source: bmcinfectdis.biomedcentral.com/articles/10.1186/s12879-024-10230-5
