SmartEBM AI Agent: A Web-based Platform for Streamlining Network Meta-Analysis via Human-AI Collaboration.

Area of Science:

• Evidence synthesis
• Comparative effectiveness research
• Health informatics

Background:

• Network meta-analysis (NMA) is crucial for comparative effectiveness research, especially in fields like complementary medicine.
• Traditional NMA workflows are labor-intensive and methodologically complex.
• Large language models offer potential for evidence synthesis but face challenges in transparency and integration.

Purpose of the Study:

• To introduce SmartEBM, a web-based platform designed for human-AI collaboration in NMA.
• To support the entire NMA workflow, from screening to evidence assessment.
• To improve the efficiency, manageability, and accessibility of NMA.

Main Methods:

• SmartEBM integrates AI-assisted functions with human oversight in a human-in-the-loop model.
• The platform features six modules for screening, data extraction, risk of bias, statistical analysis, and certainty of evidence assessment.
• It offers low-code interfaces, structured outputs, and verification-oriented workspaces for end-to-end NMA support.

Main Results:

• SmartEBM provides a unified environment for all stages of NMA.
• The platform facilitates AI-assisted tasks while maintaining researcher control over critical judgments.
• It aims to make complex evidence synthesis more manageable and traceable.

Conclusions:

• SmartEBM is a practical platform for end-to-end NMA support, not a fully automated solution.
• The platform enhances the efficiency and accessibility of NMA, particularly in complex research settings.
• It represents a significant advancement in applying AI to evidence synthesis for routine research.