Abstract
High-quality literature reviews underpin public health policy and practice. However, traditional review methods are increasingly strained by the exponential growth of scientific publications and the need for timely decision-making. Artificial intelligence (AI) tools are rapidly being adopted to support evidence synthesis, yet their appropriate use requires clearer guidance and a critical understanding of their limitations. Recent evidence further showed that AI systems can reproduce and propagate incorrect or fabricated scientific information, highlighting challenges in distinguishing valid from invalid evidence in review workflows. Improving the efficiency and transparency of evidence synthesis is critical for timely and equitable public health decision-making, particularly in rapidly evolving policy contexts, where biased or unreliable evidence can directly affect population health outcomes. This workshop is designed for early- to mid-career public health researchers and professionals involved in evidence synthesis, with basic familiarity with systematic reviews. It aims to build practical skills for the responsible use of AI in systematic and rapid reviews, focusing on two key steps: reference screening and data extraction. The workshop includes four activities: First, two experts will discuss the benefits and limitations of AI-supported approaches in these steps (10 min). Second, the audience will receive a short demonstration of selected open-access AI tools as suitable tools for reference screening and data extraction (10 min). Third, participants will engage in facilitated small-group exercises using a recently conducted systematic review without AI support as a reference standard (30 min). They will apply the introduced AI tools to screening and data extraction tasks and compare their outputs with the reference standard. This comparison will serve as the basis for a structured discussion on performance, transparency, bias, and required levels of human oversight (10 min). Together with the speakers, participants will discuss ethical considerations of AI use and its implications for research quality. This approach combines hands-on skill building with critical reflection on algorithmic bias, transparency, and methodological rigor. It supports informed decision-making on the responsible use of AI in public health reviews, emphasising the continued need for human expertise and oversight. By the end of the workshop, participants will be able to: (1) apply AI tools to support screening and data extraction within a structured review workflow; (2) compare AI-assisted outputs with manually generated results and identify discrepancies; and (3) critically assess risks, limitations, and appropriate use of AI in evidence synthesis.
Key messages
• Artificial Intelligence can support public health reviews, but only when its limitations, risks, and biases are explicitly recognised and managed.
• This workshop builds practical skills while fostering critical debate on responsible AI use in evidence synthesis.
Speakers/Panellists
Patrick Boudreau
Montreal Behavioural Medicine Centre, Concordia University, Montreal, Canada
Francesco Baglivo
University of Pisa, Pisa, Italy