AIREV: an AI-Based System to Support Systematic Reviews

The exponential growth of scientific literature has made evidence synthesis through systematic reviews essential.

Since systematic reviews are highly demanding in terms of time and resources, the use of artificial intelligence (AI) is being explored to improve the efficiency of the process. The available data are promising, but not yet conclusive.

The AIREV study (Artificial Intelligence for Scientific Reviews and Evidence Visualization) is conducted on behalf of the Epidemiology Unit of ATS – Metropolitan City of Milan, within a PNC (National Plan for Complementary Investments) funding programme of the Italian Ministry of Health. The study explores the association between air pollution and health outcomes for which the evidence is not yet well established.

Using a systematic review conducted with traditional methodology as a reference, AIREV aims to develop and evaluate a review process that relies wholly or partially on AI.

STATO PROGETTO

Ongoing

COMMITTENTE

ATS-Città metropolitana di Milano

RESPONSABILI

Pietro Dri Maria Rosa Valetto

Zadig’s activities

Zadig uses Elicit (www.elicit.com), a research and analysis tool for scientific literature based on large language models (LLMs), and develops a workflow that is similar and parallel to the traditional methodology for the selection, screening, and analysis of articles for systematic reviews.

The results produced by Elicit are compared phase by phase with those obtained using the traditional methodology.

In addition, Elicit is used to apply the criteria of the standardized and validated AMSTAR-2 EH questionnaire to the selected studies, in order to assess their methodological quality.

This approach is intended to be replicated for further systematic reviews.