AI-Powered Systematic Reviews: Transforming Health Research with Large Language Models

The landscape of health research is undergoing a rapid transformation, fueled by advancements in artificial intelligence (AI), particularly large language models (LLMs). These powerful tools are poised to revolutionize the traditionally laborious process of conducting systematic reviews, offering the potential for increased efficiency, scalability, and access to knowledge. This article explores the current state of LLM applications in systematic reviews, the methodologies being developed, and the implications for the future of evidence-based medicine.

What are Systematic Reviews and Why are They Critical?

Systematic reviews synthesize the findings of multiple studies to provide a comprehensive and unbiased assessment of a specific research question. They are considered the gold standard for evidence-based decision-making in healthcare, informing clinical guidelines, policies, and practice. Still, traditional systematic reviews are time-consuming, resource-intensive, and require significant expertise.

The Rise of LLMs in Systematic Review Methodology

Large language models, such as GPT, ChatGPT, LLaMA, Claude, Gemini, and Bard, are demonstrating remarkable capabilities in understanding and generating human-like text. Researchers are now exploring how to leverage these models to automate or assist with various stages of the systematic review process. A new methodological framework, PRISMA-DFLLM (for Domain-specific Finetuned LLMs), combines the rigor of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines with the power of LLMs.

Key Applications of LLMs in Systematic Reviews

• Search Strategy and Study Selection: LLMs can assist in developing comprehensive search strategies and screening large volumes of research articles to identify potentially relevant studies.
• Data Extraction: LLMs can be finetuned to extract key data points from included studies, such as study design, participant characteristics, and intervention details.
• Risk of Bias Assessment: While not yet fully automated, LLMs can aid in assessing the methodological quality and risk of bias of individual studies.
• Evidence Synthesis: LLMs can support summarize and synthesize the findings of included studies, identifying patterns and inconsistencies in the evidence.

PRISMA and Transparency in AI-Assisted Reviews

Maintaining transparency and methodological rigor is crucial when incorporating AI into systematic reviews. The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement provides guidelines for clear and comprehensive reporting of systematic reviews. Recognizing the unique challenges posed by AI, the PRISMA-trAIce checklist has been developed to improve the transparency and methodological integrity of AI-assisted systematic reviews.

Current Research and Findings

Recent research has focused on developing evidence tier frameworks to assess the quality of LLM-based medical studies. Studies published between January 1, 2022, and September 6, 2025, were analyzed using search terms combining general descriptors of LLMs with specific model names, focusing on original research in health-related subject areas. Researchers are also investigating the rate at which LLMs are outperforming humans in various tasks related to systematic review methodology.

Challenges and Future Directions

Despite the promising potential, several challenges remain in the implementation of LLMs for systematic reviews. These include ensuring the accuracy and reliability of LLM-generated outputs, addressing potential biases in the models, and establishing clear guidelines for responsible AI use. Future research will likely focus on developing more sophisticated LLM-assisted frameworks, improving the interpretability of LLM outputs, and exploring the use of LLMs for continuous, “living” systematic reviews that are updated as new evidence emerges.

Key Takeaways

• LLMs are transforming the field of systematic reviews, offering the potential for increased efficiency and scalability.
• The PRISMA guidelines and the PRISMA-trAIce checklist are essential for ensuring transparency and methodological rigor in AI-assisted reviews.
• Ongoing research is focused on refining LLM-assisted frameworks and addressing the challenges associated with AI implementation.
• AI-powered systematic reviews promise to accelerate the pace of evidence-based medicine and improve healthcare outcomes.