Empowering clinical trial design with agentic intelligence and real-world data

Area of Science:

• Biomedical Informatics
• Artificial Intelligence in Healthcare
• Clinical Trial Design

Background:

• Clinical trial design (CTD) is complex and requires significant expertise.
• Real-world data (RWD), like electronic health records (EHR), contains valuable practice-based evidence for CTD.
• Existing machine learning methods for extracting real-world evidence (RWE) require extensive expert collaboration.

Purpose of the Study:

• To introduce EmulatRx, an agentic framework for deriving RWE to aid CTD.
• To demonstrate EmulatRx's ability to autonomously refine trial protocols and generate actionable insights.
• To evaluate EmulatRx's effectiveness in facilitating and accelerating the CTD process.

Main Methods:

• Developed EmulatRx, an agentic framework utilizing iterative conversations and role-based agent analysis.
• Applied EmulatRx to CTD for acute diseases (septic shock, heart failure, etc.) using MIMIC-IV data.
• Applied EmulatRx to CTD for chronic diseases (Alzheimer's, Parkinson's) using the INSIGHT Network data.

Main Results:

• EmulatRx autonomously refines trial protocols through iterative agent interactions.
• The framework generates robust reports with insights to inform CTD.
• Successful application demonstrated on diverse disease types (acute and chronic) across multiple health systems.

Conclusions:

• EmulatRx effectively facilitates and accelerates the clinical trial design process.
• The agentic framework shows promise in leveraging RWE for improved CTD.
• EmulatRx represents a significant advancement in AI-driven clinical trial development.