Advocating the potential of artificial intelligence for syndrome discovery in syndromic surveillance systems: A scoping review

Abstract

Syndromic surveillance systems monitor and detect diseases in real time using existing public health data, with syndromes-sets of clinical characteristics-being central to these systems. However, defining these syndromes is a complex and time-consuming task that requires a high level of expertise and the analysis of large, often multimodal and high-dimensional datasets. This challenge is exacerbated by the emergence of new diseases, e.g., due to climate change, urbanization, or globalization. Artificial intelligence has the potential to revolutionize the process of syndrome definition in syndromic surveillance. A scoping review following the PRISMA extension guidelines was conducted to examine the role of artificial intelligence in this field. The PubMed, Google Scholar, OpenAlex, Semantic Scholar, ConnectedPapers, Web of Science, and Embase databases were reviewed. The initial search yielded 2,228 references. After removing duplicates and references outside the scope of the review, 15 studies were included. Most current solutions focus on known syndromes through supervised machine learning, primarily using English free-text data. Artificial intelligence offers immense potential, particularly through open science practices in machine learning, which can accelerate adoption and improve outcomes. However, significant challenges remain to fully realize these benefits. The review concludes with eight strategic recommendations for advancing AI-assisted syndrome discovery in public health. These recommendations address methodology, data handling, collaboration, and standardization to promote effectiveness and global applicability.

Ähnliche Arbeiten