Artificial Intelligence (AI)-Augmented “Living” Meta-Analyses toward Critical Thinking Engagement in Chemical Education and Research: A Case Study of Nanocellulose-Stabilized Pickering Emulsions

Abstract

As research on sustainable and advanced materials accelerates (e.g., cellulose-based composites and functionalized nanomaterials), research output has expanded rapidly, increasing complexity and making it challenging for industrial and engineering chemistry researchers to maintain comprehensive, up-to-date data compilation and analysis. Therefore, traditional meta-analyses, even when attempting to adhere to systematic methodologies such as the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA), face challenges with manual data curation, risking oversights and impacting reproducibility in the face of such volume. Updated meta-analysis methodologies are necessary to critically assess advances and new technologies for the upscaling processes and innovation in the advanced materials field. To address this, we propose a novel framework for creating “living” meta-analyses augmented by artificial intelligence (AI). We explore these concepts via a tutorial case study on nanocellulose-stabilized Pickering emulsions, illustrating how the integration of AI-based extraction with bibliometric mapping reveals patterns, identifies research gaps, and enables the deployment of informed decisions for future research and accelerated product development in academia and industry. Integrating Large Language Models (LLMs), such as ChatGPT and Gemini, with bibliometric platforms (e.g., ACS CAS SciFinder, Scopus, Dimensions) and VOSviewer, allowed one to systematically curate and synthesize data from over 50 publications. The resulting interactive platform reveals complex relationships among nanocellulose properties (e.g., type, modification, concentration), processing conditions, and emulsion characteristics (e.g., droplet size, stability). The database and software are available at 10.5281/zenodo.15808694. We critically discuss the current limitations of LLMs in performing meta-analyses in the chemical engineering and advanced materials fields and emphasize the role of “human-in-the-loop” expertise in interpretation.

Authors wish to acknowledge the financial support of the Spanish Ministry of Science, Innovation and Universities to the project NextPack (PID2021-124766OA-I00). Giovana Signori-Iamin received funding for her PhD thesis from the University of Girona (PhD grant IFUdG2023)

Open Access funding provided thanks to the CRUE-CSIC agreement with ACS

9